Upload SENTiVENT release artifacts

.gitattributes

@@ -137,3 +137,11 @@ data/tokens/train.jsonl filter=lfs diff=lfs merge=lfs -text
 data/uie_ere_sentence/train.jsonl filter=lfs diff=lfs merge=lfs -text
 data/uie_iabsa_sentence/train.jsonl filter=lfs diff=lfs merge=lfs -text
 data/uie_sentivent_unified_sentence/train.jsonl filter=lfs diff=lfs merge=lfs -text
+data/ere_omnievent_sentence/train.jsonl filter=lfs diff=lfs merge=lfs -text
+data/ere_oneie_sentence/train.jsonl filter=lfs diff=lfs merge=lfs -text
+data/ere_textee_sentence_e2e/train.jsonl filter=lfs diff=lfs merge=lfs -text
+data/ere_textee_sentence_eae/train.jsonl filter=lfs diff=lfs merge=lfs -text
+data/ere_textee_sentence_ed/train.jsonl filter=lfs diff=lfs merge=lfs -text
+data/sentivent_unified_bio_sentence/train.jsonl filter=lfs diff=lfs merge=lfs -text
+data/sentivent_unified_oneie_sentence/train.jsonl filter=lfs diff=lfs merge=lfs -text
+data/sentivent_unified_uie_sentence/train.jsonl filter=lfs diff=lfs merge=lfs -text

README.md

@@ -72,454 +72,102 @@ configs:
     path: data/sentivent_unified_document_with_canonical_referents/validation.parquet
   - split: test
     path: data/sentivent_unified_document_with_canonical_referents/test.parquet
-- config_name: documents
+- config_name: sentivent_unified_bio_sentence
   data_files:
   - split: train
-    path: data/documents/train.parquet
+    path: data/sentivent_unified_bio_sentence/train.parquet
   - split: validation
-    path: data/documents/validation.parquet
+    path: data/sentivent_unified_bio_sentence/validation.parquet
   - split: test
-    path: data/documents/test.parquet
-- config_name: sentences
+    path: data/sentivent_unified_bio_sentence/test.parquet
+- config_name: ere_textee_sentence_e2e
   data_files:
   - split: train
-    path: data/sentences/train.parquet
+    path: data/ere_textee_sentence_e2e/train.parquet
   - split: validation
-    path: data/sentences/validation.parquet
+    path: data/ere_textee_sentence_e2e/validation.parquet
   - split: test
-    path: data/sentences/test.parquet
-- config_name: tokens
+    path: data/ere_textee_sentence_e2e/test.parquet
+- config_name: ere_textee_sentence_eae
   data_files:
   - split: train
-    path: data/tokens/train.parquet
+    path: data/ere_textee_sentence_eae/train.parquet
   - split: validation
-    path: data/tokens/validation.parquet
+    path: data/ere_textee_sentence_eae/validation.parquet
   - split: test
-    path: data/tokens/test.parquet
-- config_name: participants
+    path: data/ere_textee_sentence_eae/test.parquet
+- config_name: ere_textee_sentence_ed
   data_files:
   - split: train
-    path: data/participants/train.parquet
+    path: data/ere_textee_sentence_ed/train.parquet
   - split: validation
-    path: data/participants/validation.parquet
+    path: data/ere_textee_sentence_ed/validation.parquet
   - split: test
-    path: data/participants/test.parquet
-- config_name: events
+    path: data/ere_textee_sentence_ed/test.parquet
+- config_name: iabsa_asqp_sentence
   data_files:
   - split: train
-    path: data/events/train.parquet
+    path: data/iabsa_asqp_sentence/train.parquet
   - split: validation
-    path: data/events/validation.parquet
+    path: data/iabsa_asqp_sentence/validation.parquet
   - split: test
-    path: data/events/test.parquet
-- config_name: event_arguments
+    path: data/iabsa_asqp_sentence/test.parquet
+- config_name: iabsa_aste_sentence
   data_files:
   - split: train
-    path: data/event_arguments/train.parquet
+    path: data/iabsa_aste_sentence/train.parquet
   - split: validation
-    path: data/event_arguments/validation.parquet
+    path: data/iabsa_aste_sentence/validation.parquet
   - split: test
-    path: data/event_arguments/test.parquet
-- config_name: sentiment
+    path: data/iabsa_aste_sentence/test.parquet
+- config_name: iabsa_quad_sentence
   data_files:
   - split: train
-    path: data/sentiment/train.parquet
+    path: data/iabsa_quad_sentence/train.parquet
   - split: validation
-    path: data/sentiment/validation.parquet
+    path: data/iabsa_quad_sentence/validation.parquet
   - split: test
-    path: data/sentiment/test.parquet
-- config_name: sentiment_targets
+    path: data/iabsa_quad_sentence/test.parquet
+- config_name: ere_dygiepp_document
   data_files:
   - split: train
-    path: data/sentiment_targets/train.parquet
+    path: data/ere_dygiepp_document/train.parquet
   - split: validation
-    path: data/sentiment_targets/validation.parquet
+    path: data/ere_dygiepp_document/validation.parquet
   - split: test
-    path: data/sentiment_targets/test.parquet
-- config_name: event_sentiment
+    path: data/ere_dygiepp_document/test.parquet
+- config_name: sentivent_unified_uie_sentence
   data_files:
   - split: train
-    path: data/event_sentiment/train.parquet
+    path: data/sentivent_unified_uie_sentence/train.parquet
   - split: validation
-    path: data/event_sentiment/validation.parquet
+    path: data/sentivent_unified_uie_sentence/validation.parquet
   - split: test
-    path: data/event_sentiment/test.parquet
-- config_name: relations
+    path: data/sentivent_unified_uie_sentence/test.parquet
+- config_name: ere_omnievent_sentence
   data_files:
   - split: train
-    path: data/relations/train.parquet
+    path: data/ere_omnievent_sentence/train.parquet
   - split: validation
-    path: data/relations/validation.parquet
+    path: data/ere_omnievent_sentence/validation.parquet
   - split: test
-    path: data/relations/test.parquet
-- config_name: task_ere_document
+    path: data/ere_omnievent_sentence/test.parquet
+- config_name: sentivent_unified_oneie_sentence
   data_files:
   - split: train
-    path: data/task_ere_document/train.parquet
+    path: data/sentivent_unified_oneie_sentence/train.parquet
   - split: validation
-    path: data/task_ere_document/validation.parquet
+    path: data/sentivent_unified_oneie_sentence/validation.parquet
   - split: test
-    path: data/task_ere_document/test.parquet
-- config_name: task_ere_sentence
+    path: data/sentivent_unified_oneie_sentence/test.parquet
+- config_name: ere_oneie_sentence
   data_files:
   - split: train
-    path: data/task_ere_sentence/train.parquet
+    path: data/ere_oneie_sentence/train.parquet
   - split: validation
-    path: data/task_ere_sentence/validation.parquet
+    path: data/ere_oneie_sentence/validation.parquet
   - split: test
-    path: data/task_ere_sentence/test.parquet
-- config_name: task_ere_token
-  data_files:
-  - split: train
-    path: data/task_ere_token/train.parquet
-  - split: validation
-    path: data/task_ere_token/validation.parquet
-  - split: test
-    path: data/task_ere_token/test.parquet
-- config_name: task_iabsa_document
-  data_files:
-  - split: train
-    path: data/task_iabsa_document/train.parquet
-  - split: validation
-    path: data/task_iabsa_document/validation.parquet
-  - split: test
-    path: data/task_iabsa_document/test.parquet
-- config_name: task_iabsa_sentence
-  data_files:
-  - split: train
-    path: data/task_iabsa_sentence/train.parquet
-  - split: validation
-    path: data/task_iabsa_sentence/validation.parquet
-  - split: test
-    path: data/task_iabsa_sentence/test.parquet
-- config_name: task_iabsa_token
-  data_files:
-  - split: train
-    path: data/task_iabsa_token/train.parquet
-  - split: validation
-    path: data/task_iabsa_token/validation.parquet
-  - split: test
-    path: data/task_iabsa_token/test.parquet
-- config_name: task_sentivent_unified_document
-  data_files:
-  - split: train
-    path: data/task_sentivent_unified_document/train.parquet
-  - split: validation
-    path: data/task_sentivent_unified_document/validation.parquet
-  - split: test
-    path: data/task_sentivent_unified_document/test.parquet
-- config_name: task_sentivent_unified_sentence
-  data_files:
-  - split: train
-    path: data/task_sentivent_unified_sentence/train.parquet
-  - split: validation
-    path: data/task_sentivent_unified_sentence/validation.parquet
-  - split: test
-    path: data/task_sentivent_unified_sentence/test.parquet
-- config_name: task_sentivent_unified_token
-  data_files:
-  - split: train
-    path: data/task_sentivent_unified_token/train.parquet
-  - split: validation
-    path: data/task_sentivent_unified_token/validation.parquet
-  - split: test
-    path: data/task_sentivent_unified_token/test.parquet
-- config_name: llm_schema_ere_document
-  data_files:
-  - split: train
-    path: data/llm_schema_ere_document/train.parquet
-  - split: validation
-    path: data/llm_schema_ere_document/validation.parquet
-  - split: test
-    path: data/llm_schema_ere_document/test.parquet
-- config_name: llm_schema_ere_sentence
-  data_files:
-  - split: train
-    path: data/llm_schema_ere_sentence/train.parquet
-  - split: validation
-    path: data/llm_schema_ere_sentence/validation.parquet
-  - split: test
-    path: data/llm_schema_ere_sentence/test.parquet
-- config_name: llm_schema_iabsa_document
-  data_files:
-  - split: train
-    path: data/llm_schema_iabsa_document/train.parquet
-  - split: validation
-    path: data/llm_schema_iabsa_document/validation.parquet
-  - split: test
-    path: data/llm_schema_iabsa_document/test.parquet
-- config_name: llm_schema_iabsa_sentence
-  data_files:
-  - split: train
-    path: data/llm_schema_iabsa_sentence/train.parquet
-  - split: validation
-    path: data/llm_schema_iabsa_sentence/validation.parquet
-  - split: test
-    path: data/llm_schema_iabsa_sentence/test.parquet
-- config_name: llm_schema_sentivent_unified_document
-  data_files:
-  - split: train
-    path: data/llm_schema_sentivent_unified_document/train.parquet
-  - split: validation
-    path: data/llm_schema_sentivent_unified_document/validation.parquet
-  - split: test
-    path: data/llm_schema_sentivent_unified_document/test.parquet
-- config_name: llm_schema_sentivent_unified_sentence
-  data_files:
-  - split: train
-    path: data/llm_schema_sentivent_unified_sentence/train.parquet
-  - split: validation
-    path: data/llm_schema_sentivent_unified_sentence/validation.parquet
-  - split: test
-    path: data/llm_schema_sentivent_unified_sentence/test.parquet
-- config_name: token_label_event_argument_role_bio
-  data_files:
-  - split: train
-    path: data/token_label_event_argument_role_bio/train.parquet
-  - split: validation
-    path: data/token_label_event_argument_role_bio/validation.parquet
-  - split: test
-    path: data/token_label_event_argument_role_bio/test.parquet
-- config_name: token_label_event_trigger_bio
-  data_files:
-  - split: train
-    path: data/token_label_event_trigger_bio/train.parquet
-  - split: validation
-    path: data/token_label_event_trigger_bio/validation.parquet
-  - split: test
-    path: data/token_label_event_trigger_bio/test.parquet
-- config_name: token_label_event_type_bio
-  data_files:
-  - split: train
-    path: data/token_label_event_type_bio/train.parquet
-  - split: validation
-    path: data/token_label_event_type_bio/validation.parquet
-  - split: test
-    path: data/token_label_event_type_bio/test.parquet
-- config_name: token_label_sentiment_expression_bio
-  data_files:
-  - split: train
-    path: data/token_label_sentiment_expression_bio/train.parquet
-  - split: validation
-    path: data/token_label_sentiment_expression_bio/validation.parquet
-  - split: test
-    path: data/token_label_sentiment_expression_bio/test.parquet
-- config_name: token_label_sentiment_target_bio
-  data_files:
-  - split: train
-    path: data/token_label_sentiment_target_bio/train.parquet
-  - split: validation
-    path: data/token_label_sentiment_target_bio/validation.parquet
-  - split: test
-    path: data/token_label_sentiment_target_bio/test.parquet
-- config_name: token_label_sentivent_unified_bio
-  data_files:
-  - split: train
-    path: data/token_label_sentivent_unified_bio/train.parquet
-  - split: validation
-    path: data/token_label_sentivent_unified_bio/validation.parquet
-  - split: test
-    path: data/token_label_sentivent_unified_bio/test.parquet
-- config_name: textee_e2e_document
-  data_files:
-  - split: train
-    path: data/textee_e2e_document/train.parquet
-  - split: validation
-    path: data/textee_e2e_document/validation.parquet
-  - split: test
-    path: data/textee_e2e_document/test.parquet
-- config_name: textee_e2e_sentence
-  data_files:
-  - split: train
-    path: data/textee_e2e_sentence/train.parquet
-  - split: validation
-    path: data/textee_e2e_sentence/validation.parquet
-  - split: test
-    path: data/textee_e2e_sentence/test.parquet
-- config_name: textee_eae_document
-  data_files:
-  - split: train
-    path: data/textee_eae_document/train.parquet
-  - split: validation
-    path: data/textee_eae_document/validation.parquet
-  - split: test
-    path: data/textee_eae_document/test.parquet
-- config_name: textee_eae_sentence
-  data_files:
-  - split: train
-    path: data/textee_eae_sentence/train.parquet
-  - split: validation
-    path: data/textee_eae_sentence/validation.parquet
-  - split: test
-    path: data/textee_eae_sentence/test.parquet
-- config_name: textee_ed_document
-  data_files:
-  - split: train
-    path: data/textee_ed_document/train.parquet
-  - split: validation
-    path: data/textee_ed_document/validation.parquet
-  - split: test
-    path: data/textee_ed_document/test.parquet
-- config_name: textee_ed_sentence
-  data_files:
-  - split: train
-    path: data/textee_ed_sentence/train.parquet
-  - split: validation
-    path: data/textee_ed_sentence/validation.parquet
-  - split: test
-    path: data/textee_ed_sentence/test.parquet
-- config_name: absa_asqp_quad_line
-  data_files:
-  - split: train
-    path: data/absa_asqp_quad_line/train.parquet
-  - split: validation
-    path: data/absa_asqp_quad_line/validation.parquet
-  - split: test
-    path: data/absa_asqp_quad_line/test.parquet
-- config_name: absa_aste_triplet_line
-  data_files:
-  - split: train
-    path: data/absa_aste_triplet_line/train.parquet
-  - split: validation
-    path: data/absa_aste_triplet_line/validation.parquet
-  - split: test
-    path: data/absa_aste_triplet_line/test.parquet
-- config_name: absa_gts_event_sentiment
-  data_files:
-  - split: train
-    path: data/absa_gts_event_sentiment/train.parquet
-  - split: validation
-    path: data/absa_gts_event_sentiment/validation.parquet
-  - split: test
-    path: data/absa_gts_event_sentiment/test.parquet
-- config_name: absa_gts_sentiment
-  data_files:
-  - split: train
-    path: data/absa_gts_sentiment/train.parquet
-  - split: validation
-    path: data/absa_gts_sentiment/validation.parquet
-  - split: test
-    path: data/absa_gts_sentiment/test.parquet
-- config_name: absa_iabsa_quad
-  data_files:
-  - split: train
-    path: data/absa_iabsa_quad/train.parquet
-  - split: validation
-    path: data/absa_iabsa_quad/validation.parquet
-  - split: test
-    path: data/absa_iabsa_quad/test.parquet
-- config_name: absa_iabsa_setfit
-  data_files:
-  - split: train
-    path: data/absa_iabsa_setfit/train.parquet
-  - split: validation
-    path: data/absa_iabsa_setfit/validation.parquet
-  - split: test
-    path: data/absa_iabsa_setfit/test.parquet
-- config_name: absa_iabsa_triplet
-  data_files:
-  - split: train
-    path: data/absa_iabsa_triplet/train.parquet
-  - split: validation
-    path: data/absa_iabsa_triplet/validation.parquet
-  - split: test
-    path: data/absa_iabsa_triplet/test.parquet
-- config_name: absa_implicit_polarity
-  data_files:
-  - split: train
-    path: data/absa_implicit_polarity/train.parquet
-  - split: validation
-    path: data/absa_implicit_polarity/validation.parquet
-  - split: test
-    path: data/absa_implicit_polarity/test.parquet
-- config_name: absa_sentivent_unified_quad
-  data_files:
-  - split: train
-    path: data/absa_sentivent_unified_quad/train.parquet
-  - split: validation
-    path: data/absa_sentivent_unified_quad/validation.parquet
-  - split: test
-    path: data/absa_sentivent_unified_quad/test.parquet
-- config_name: dygiepp_events
-  data_files:
-  - split: train
-    path: data/dygiepp_events/train.parquet
-  - split: validation
-    path: data/dygiepp_events/validation.parquet
-  - split: test
-    path: data/dygiepp_events/test.parquet
-- config_name: uie_ere_sentence
-  data_files:
-  - split: train
-    path: data/uie_ere_sentence/train.parquet
-  - split: validation
-    path: data/uie_ere_sentence/validation.parquet
-  - split: test
-    path: data/uie_ere_sentence/test.parquet
-- config_name: uie_iabsa_sentence
-  data_files:
-  - split: train
-    path: data/uie_iabsa_sentence/train.parquet
-  - split: validation
-    path: data/uie_iabsa_sentence/validation.parquet
-  - split: test
-    path: data/uie_iabsa_sentence/test.parquet
-- config_name: uie_sentivent_unified_sentence
-  data_files:
-  - split: train
-    path: data/uie_sentivent_unified_sentence/train.parquet
-  - split: validation
-    path: data/uie_sentivent_unified_sentence/validation.parquet
-  - split: test
-    path: data/uie_sentivent_unified_sentence/test.parquet
-- config_name: legacy_ie_dygiepp_event_sentiment
-  data_files:
-  - split: train
-    path: data/legacy_ie_dygiepp_event_sentiment/train.parquet
-  - split: validation
-    path: data/legacy_ie_dygiepp_event_sentiment/validation.parquet
-  - split: test
-    path: data/legacy_ie_dygiepp_event_sentiment/test.parquet
-- config_name: legacy_ie_dygiepp_events
-  data_files:
-  - split: train
-    path: data/legacy_ie_dygiepp_events/train.parquet
-  - split: validation
-    path: data/legacy_ie_dygiepp_events/validation.parquet
-  - split: test
-    path: data/legacy_ie_dygiepp_events/test.parquet
-- config_name: legacy_ie_omnievent_event_sentiment
-  data_files:
-  - split: train
-    path: data/legacy_ie_omnievent_event_sentiment/train.parquet
-  - split: validation
-    path: data/legacy_ie_omnievent_event_sentiment/validation.parquet
-  - split: test
-    path: data/legacy_ie_omnievent_event_sentiment/test.parquet
-- config_name: legacy_ie_omnievent_events
-  data_files:
-  - split: train
-    path: data/legacy_ie_omnievent_events/train.parquet
-  - split: validation
-    path: data/legacy_ie_omnievent_events/validation.parquet
-  - split: test
-    path: data/legacy_ie_omnievent_events/test.parquet
-- config_name: legacy_ie_oneie_event_sentiment
-  data_files:
-  - split: train
-    path: data/legacy_ie_oneie_event_sentiment/train.parquet
-  - split: validation
-    path: data/legacy_ie_oneie_event_sentiment/validation.parquet
-  - split: test
-    path: data/legacy_ie_oneie_event_sentiment/test.parquet
-- config_name: legacy_ie_oneie_events
-  data_files:
-  - split: train
-    path: data/legacy_ie_oneie_events/train.parquet
-  - split: validation
-    path: data/legacy_ie_oneie_events/validation.parquet
-  - split: test
-    path: data/legacy_ie_oneie_events/test.parquet
+    path: data/ere_oneie_sentence/test.parquet
 ---
 
 # SENTiVENT v1.1 Economic Events and Implicit Sentiment
@@ -538,7 +186,7 @@ SENTiVENT is intended for reproducible evaluation of information extraction and
 - Human-annotated economic event, argument, sentiment-expression, sentiment-target, and event-sentiment labels.
 - A joint document-level target that links event triggers, participants, fillers, sentiment expressions, entity/aspect/event targets, and polarity.
 - Consistent train/dev/test splits across configs matching the pre-existing SENTiVENT papers for benchmark comparison.
-- Flattened table configs and research-standard document, sentence, token, and token-span task views for corpus inspection, baselines, prompting, and evaluation.
+- Research-standard document, sentence, and token-span interoperability views for baselines, prompting, and evaluation.
 
 Implicit sentiment captures cases where factual business-news content changes a reader-investor's attitude toward a company, asset, person, product, or event because readers know what is desirable or undesirable in financial markets, even without words like `great` or `bad`.
 Implicit sentiment in financial news is challenging because articles are written objectively and the evidence is lexically diverse and requires common-sense world knowledge to resolve.
@@ -554,12 +202,13 @@ from datasets import load_dataset
 
 unified = load_dataset("GillesJacobs/sentivent")
 unified_continuous = load_dataset("GillesJacobs/sentivent", "sentivent_unified_document_continuous")
-unified_tokens = load_dataset("GillesJacobs/sentivent", "task_sentivent_unified_token")
-unified_sentences = load_dataset("GillesJacobs/sentivent", "task_sentivent_unified_sentence")
-documents = load_dataset("GillesJacobs/sentivent", "documents")
-events = load_dataset("GillesJacobs/sentivent", "events")
+uie_sentences = load_dataset("GillesJacobs/sentivent", "sentivent_unified_uie_sentence")
+textee_events = load_dataset("GillesJacobs/sentivent", "ere_textee_sentence_e2e")
+iabsa_quads = load_dataset("GillesJacobs/sentivent", "iabsa_quad_sentence")
 ```
 
+The anonymized raw WebAnno/UIMA XMI source export is included under `raw_webanno_export/` for users who need the original annotation files. It is a source artifact, not a recommended modeling or evaluation config.
+
 ## Main Task
 
 - The primary public config is `sentivent_unified_document`, the compact view for joint economic event extraction, implicit economic sentiment, and event-targeted sentiment in one structured target.
@@ -699,81 +348,32 @@ The holdout test set is a gold-standard expert-adjudicated reference set from a
 | --- | --- | --- |
 | sentivent_unified_document | Canonical joint benchmark for ML and LLM evaluation | Full-article structured IE: extract economic event triggers, types/subtypes, participants, fillers, implicit sentiment expressions/targets, and event-sentiment links. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554), and [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). |
 | sentivent_unified_document_continuous | Systems that cannot represent discontinuous spans | Same joint document task for span-limited extractive tools; discontinuous spans are filled to one covering extent, so this is a lossy convenience view. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4) and [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). |
-| task_sentivent_unified_token | Token-span baselines and sequence-model evaluation | Token-sequence joint extraction: predict trigger, argument, sentiment-expression, target, and link spans from sentence tokens and offsets. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4) and [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). |
-| task_sentivent_unified_sentence | Sentence-scoped structured-output prompting | Sentence-level structured-output prompt/eval rows: each row carries an instruction, sentence text, and scoped event, sentiment, and link gold. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4) and [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). |
-| documents | Corpus browsing, joins, and metadata checks | Normalized article table for inspecting full text, split membership, and annotation counts before joining flatter event and sentiment tables. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554), and [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). |
-| events | Event-only extraction and label analysis | Canonical event table for ACE/ERE-style economic event extraction: one row per trigger with type/subtype labels, spans, and argument ids. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). |
+| sentivent_unified_document_with_canonical_referents | Auditing canonical-reference resolution | Same joint document task plus inline canonical-reference evidence for resolved arguments and targets. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554), and [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). |
 
 ## Dataset Configs
 
 - `sentivent_unified_document` is the default and recommended config for the full joint extraction task over article text.
 - `sentivent_unified_document_continuous` is a derived evaluation variant that fills discontinuous spans to continuous ranges.
-- `task_sentivent_unified_document`, `task_sentivent_unified_sentence`, and `task_sentivent_unified_token` provide secondary task-row views with structured gold JSON.
-- `task_ere_document`, `task_ere_sentence`, and `task_ere_token` expose the event extraction labels.
-- `task_iabsa_document`, `task_iabsa_sentence`, and `task_iabsa_token` expose implicit aspect-based economic sentiment labels.
-- Canonical configs such as `documents`, `sentences`, `tokens`, `events`, `event_arguments`, `sentiment`, `sentiment_targets`, and `event_sentiment` expose the annotation layers as tables.
+- `sentivent_unified_document_with_canonical_referents` is an opt-in variant that keeps canonical-reference evidence inline.
+- Interoperability configs expose complete task views for common event extraction, implicit ABSA, UIE, OneIE, OmniEvent, DyGIE++, TextEE, and BIO pipelines.
 
 | Config | Description | train / validation / test |
 | --- | --- | --- |
 | sentivent_unified_document | Primary joint document-level benchmark: extract economic events, participants, fillers, implicit sentiment expressions, entity/aspect/event targets, polarity, and event-sentiment links. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554), and [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). | 228 / 30 / 30 |
 | sentivent_unified_document_continuous | Lossy continuous-span variant of the joint document extraction task for tools that cannot represent discontinuous spans. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4) and [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 228 / 30 / 30 |
 | sentivent_unified_document_with_canonical_referents | Opt-in joint document config with the same resolved spans as the default view plus inline `resolved_from` evidence for canonical referents, including standalone canonical-reference annotation items when needed. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554), and [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). | 228 / 30 / 30 |
-| documents | Article-level corpus table for full text, split membership, and annotation counts before joining other tables. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554), and [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). | 228 / 30 / 30 |
-| sentences | Sentence segmentation table for sentence-scoped tasks and span joins. See the [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). | 5475 / 681 / 727 |
-| tokens | Token table for token-span alignment, BIO labeling, and offset validation. See the [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). | 135317 / 17955 / 17126 |
-| participants | Entity, participant, and filler mention table used as event arguments and sentiment targets. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 11188 / 1459 / 2098 |
-| events | Economic event extraction table: one row per event trigger with type/subtype, attributes, spans, and argument ids. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 4636 / 625 / 984 |
-| event_arguments | Event argument edge table linking events to participant or filler spans with roles. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 10135 / 1346 / 2299 |
-| sentiment | Implicit economic sentiment expression table with polarity and target ids. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 2954 / 387 / 328 |
-| sentiment_targets | Sentiment-target edge table linking sentiment expressions to entity/aspect or event targets. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 3531 / 483 / 415 |
-| event_sentiment | Joint event-sentiment relation table for sentiment implied by or attached to events. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 71 / 11 / 20 |
-| relations | Normalized relation table for cross-layer links such as event sentiment and coreference. See the [SENTiVENT thesis](https://biblio.ugent.be/publication/8728891). | 1710 / 200 / 397 |
-| task_ere_document | Document-level economic event extraction prompt/eval rows: extract event triggers, types/subtypes, participants, and fillers. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 228 / 30 / 30 |
-| task_ere_sentence | Sentence-level economic event extraction prompt/eval rows: extract event triggers, types/subtypes, participants, and fillers. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
-| task_ere_token | Token-span-level economic event extraction prompt/eval rows: extract event triggers, types/subtypes, participants, and fillers. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
-| task_iabsa_document | Document-level implicit economic ABSA prompt/eval rows: extract sentiment expressions, entity/aspect targets, and polarity. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 228 / 30 / 30 |
-| task_iabsa_sentence | Sentence-level implicit economic ABSA prompt/eval rows: extract sentiment expressions, entity/aspect targets, and polarity. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| task_iabsa_token | Token-span-level implicit economic ABSA prompt/eval rows: extract sentiment expressions, entity/aspect targets, and polarity. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| task_sentivent_unified_document | Document-level joint structured-output rows combining economic event extraction, implicit sentiment, and event-sentiment links. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4) and [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 228 / 30 / 30 |
-| task_sentivent_unified_sentence | Sentence-level joint structured-output rows combining economic event extraction, implicit sentiment, and event-sentiment links. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4) and [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| task_sentivent_unified_token | Token-span-level joint structured-output rows combining economic event extraction, implicit sentiment, and event-sentiment links. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4) and [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| llm_schema_ere_document | Schema-constrained LLM evaluation row for economic event extraction: chat messages, JSON Schema target, expected JSON, and GoLLIE-style class prompt sidecars. See [GoLLIE](https://github.com/hitz-zentroa/GoLLIE), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 228 / 30 / 30 |
-| llm_schema_ere_sentence | Schema-constrained LLM evaluation row for economic event extraction: chat messages, JSON Schema target, expected JSON, and GoLLIE-style class prompt sidecars. See [GoLLIE](https://github.com/hitz-zentroa/GoLLIE), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| llm_schema_iabsa_document | Schema-constrained LLM evaluation row for implicit economic ABSA: chat messages, JSON Schema target, expected JSON, and GoLLIE-style class prompt sidecars. See [GoLLIE](https://github.com/hitz-zentroa/GoLLIE), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 228 / 30 / 30 |
-| llm_schema_iabsa_sentence | Schema-constrained LLM evaluation row for implicit economic ABSA: chat messages, JSON Schema target, expected JSON, and GoLLIE-style class prompt sidecars. See [GoLLIE](https://github.com/hitz-zentroa/GoLLIE), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| llm_schema_sentivent_unified_document | Schema-constrained LLM evaluation row for joint event and implicit sentiment extraction: chat messages, JSON Schema target, expected JSON, and GoLLIE-style class prompt sidecars. See [GoLLIE](https://github.com/hitz-zentroa/GoLLIE), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 228 / 30 / 30 |
-| llm_schema_sentivent_unified_sentence | Schema-constrained LLM evaluation row for joint event and implicit sentiment extraction: chat messages, JSON Schema target, expected JSON, and GoLLIE-style class prompt sidecars. See [GoLLIE](https://github.com/hitz-zentroa/GoLLIE), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| token_label_event_argument_role_bio | Event argument role sequence labeling: one BIO tag per token, with `ner_tags` for Hugging Face token-classification code and sidecars for overlaps. See [HF token classification](https://huggingface.co/tasks/token-classification), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| token_label_event_trigger_bio | Event trigger sequence labeling: one BIO tag per token, with `ner_tags` for Hugging Face token-classification code and sidecars for overlaps. See [HF token classification](https://huggingface.co/tasks/token-classification), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| token_label_event_type_bio | Event trigger type sequence labeling: one BIO tag per token, with `ner_tags` for Hugging Face token-classification code and sidecars for overlaps. See [HF token classification](https://huggingface.co/tasks/token-classification), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| token_label_sentiment_expression_bio | Sentiment-expression sequence labeling: one BIO tag per token, with `ner_tags` for Hugging Face token-classification code and sidecars for overlaps. See [HF token classification](https://huggingface.co/tasks/token-classification), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| token_label_sentiment_target_bio | Sentiment-target sequence labeling: one BIO tag per token, with `ner_tags` for Hugging Face token-classification code and sidecars for overlaps. See [HF token classification](https://huggingface.co/tasks/token-classification), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| token_label_sentivent_unified_bio | Joint event and sentiment sequence labeling: one BIO tag per token, with `ner_tags` for Hugging Face token-classification code and sidecars for overlaps. See [HF token classification](https://huggingface.co/tasks/token-classification), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| textee_e2e_document | TextEE-style document-window extension rows for event detection plus argument extraction; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 228 / 30 / 30 |
-| textee_e2e_sentence | TextEE-style sentence/window rows for event detection plus argument extraction; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
-| textee_eae_document | TextEE-style document-window extension rows for event argument extraction; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 228 / 30 / 30 |
-| textee_eae_sentence | TextEE-style sentence/window rows for event argument extraction; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
-| textee_ed_document | TextEE-style document-window extension rows for event trigger detection/classification; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 228 / 30 / 30 |
-| textee_ed_sentence | TextEE-style sentence/window rows for event trigger detection/classification; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
-| absa_asqp_quad_line | ASQP-style aspect sentiment quad prediction as delimiter-line rows with character spans and 0/1/2 polarity ids. See [ASQP](https://github.com/Datastory-CN/ASQP-Datasets) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| absa_aste_triplet_line | ASTE-style aspect sentiment triplet extraction as delimiter-line rows and companion `.txt` split files. See [ASTE](https://github.com/xuuuluuu/SemEval-Triplet-data) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| absa_gts_event_sentiment | Legacy SENTiVENT GTS-style event-sentiment triplet rows for historical comparison; not the full GTS matrix format. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| absa_gts_sentiment | Legacy SENTiVENT GTS-style sentiment triplet rows for implicit ABSA comparison. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| absa_iabsa_quad | Implicit economic ABSA quad extraction: target, aspect category, opinion expression, and polarity. See [ASQP](https://github.com/Datastory-CN/ASQP-Datasets) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| absa_iabsa_setfit | SetFit-compatible aspect polarity classification rows for implicit economic sentiment targets. See [SetFit ABSA](https://huggingface.co/docs/setfit/en/how_to/absa) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 2200 / 306 / 266 |
-| absa_iabsa_triplet | Implicit economic ABSA triplet extraction: target, opinion expression, and polarity. See [ASTE](https://github.com/xuuuluuu/SemEval-Triplet-data) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| absa_implicit_polarity | Legacy implicit-polarity classification rows from the SENTiVENT sentiment experiments. See the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 7336 / 986 / 1234 |
-| absa_sentivent_unified_quad | SENTiVENT joint quad view that keeps event-targeted sentiment and event-sentiment link ids. See the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4) and [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| dygiepp_events | Upstream DyGIE++ document-level event extraction rows with document-token indices, NER spans, and event argument records. See [DyGIE++](https://github.com/dwadden/dygiepp/blob/master/doc/data.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 228 / 30 / 30 |
-| uie_ere_sentence | UIE text-to-spot-asoc sentence rows for economic event extraction as trigger spots with argument associations. See [UIE](https://github.com/universal-ie/UIE/blob/main/docs/DATASETS.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
-| uie_iabsa_sentence | UIE text-to-spot-asoc sentence rows for implicit economic ABSA as sentiment-expression spots with target/polarity associations. See [UIE](https://github.com/universal-ie/UIE/blob/main/docs/DATASETS.md) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| uie_sentivent_unified_sentence | UIE text-to-spot-asoc sentence rows for joint event and sentiment extraction as UIE spot-association targets. See [UIE](https://github.com/universal-ie/UIE/blob/main/docs/DATASETS.md), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
-| legacy_ie_dygiepp_event_sentiment | legacy DyGIE++-style event rows with SENTiVENT sentiment-target relation extensions; intended for legacy parity or framework adapters. See [DyGIE++](https://github.com/dwadden/dygiepp/blob/master/doc/data.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 228 / 30 / 30 |
-| legacy_ie_dygiepp_events | legacy DyGIE++-style event rows; intended for legacy parity or framework adapters. See [DyGIE++](https://github.com/dwadden/dygiepp/blob/master/doc/data.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 228 / 30 / 30 |
-| legacy_ie_omnievent_event_sentiment | OmniEvent unified event/entity rows with SENTiVENT sentiment-target relation extensions; intended for legacy parity or framework adapters. See [OmniEvent](https://github.com/THU-KEG/OmniEvent/tree/main/scripts/data_processing) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5474 / 681 / 727 |
-| legacy_ie_omnievent_events | OmniEvent unified event/entity rows; intended for legacy parity or framework adapters. See [OmniEvent](https://github.com/THU-KEG/OmniEvent/tree/main/scripts/data_processing) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5474 / 681 / 727 |
-| legacy_ie_oneie_event_sentiment | OneIE/TextEE-style event rows with SENTiVENT sentiment-target relation extensions; intended for legacy parity or framework adapters. See [OneIE](https://aclanthology.org/2020.acl-main.713/), [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md), and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5474 / 681 / 727 |
-| legacy_ie_oneie_events | OneIE/TextEE-style event rows; intended for legacy parity or framework adapters. See [OneIE](https://aclanthology.org/2020.acl-main.713/), [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md), and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5474 / 681 / 727 |
+| sentivent_unified_bio_sentence | Joint event and sentiment sequence labeling: one BIO tag per token, with `ner_tags` for Hugging Face token-classification code and sidecars for overlaps. See [HF token classification](https://huggingface.co/tasks/token-classification), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
+| ere_textee_sentence_e2e | TextEE-style sentence/window rows for event detection plus argument extraction; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
+| ere_textee_sentence_eae | TextEE-style sentence/window rows for event argument extraction; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
+| ere_textee_sentence_ed | TextEE-style sentence/window rows for event trigger detection/classification; sentence configs follow the processed JSONL keys used by TextEE. See [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5475 / 681 / 727 |
+| iabsa_asqp_sentence | ASQP-style aspect sentiment quad prediction as delimiter-line rows with character spans and 0/1/2 polarity ids. See [ASQP](https://github.com/Datastory-CN/ASQP-Datasets) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
+| iabsa_aste_sentence | ASTE-style aspect sentiment triplet extraction as delimiter-line rows and companion `.txt` split files. See [ASTE](https://github.com/xuuuluuu/SemEval-Triplet-data) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
+| iabsa_quad_sentence | Implicit economic ABSA quad extraction: target, aspect category, opinion expression, and polarity. See [ASQP](https://github.com/Datastory-CN/ASQP-Datasets) and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
+| ere_dygiepp_document | Upstream DyGIE++ document-level event extraction rows with document-token indices, NER spans, and event argument records. See [DyGIE++](https://github.com/dwadden/dygiepp/blob/master/doc/data.md) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 228 / 30 / 30 |
+| sentivent_unified_uie_sentence | UIE text-to-spot-asoc sentence rows for joint event and sentiment extraction as UIE spot-association targets. See [UIE](https://github.com/universal-ie/UIE/blob/main/docs/DATASETS.md), the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4), and the [implicit sentiment paper](https://www.mdpi.com/2079-9292/10/20/2554). | 5475 / 681 / 727 |
+| ere_omnievent_sentence | OmniEvent unified event/entity rows; intended for legacy parity or framework adapters. See [OmniEvent](https://github.com/THU-KEG/OmniEvent/tree/main/scripts/data_processing) and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5474 / 681 / 727 |
+| sentivent_unified_oneie_sentence | OneIE/TextEE-style event rows with SENTiVENT sentiment-target relation extensions; intended for legacy parity or framework adapters. See [OneIE](https://aclanthology.org/2020.acl-main.713/), [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md), and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5474 / 681 / 727 |
+| ere_oneie_sentence | OneIE/TextEE-style event rows; intended for legacy parity or framework adapters. See [OneIE](https://aclanthology.org/2020.acl-main.713/), [TextEE](https://github.com/ej0cl6/TextEE/blob/main/data/README.md), and the [event extraction paper](https://doi.org/10.1007/s10579-021-09562-4). | 5474 / 681 / 727 |
 
 ## Splits
 
@@ -830,8 +430,8 @@ SENTiVENT is released under the Creative Commons Attribution-NonCommercial 4.0 I
 
 ## Release Provenance
 
-- Release tag: `sentivent-guidelines-v1.1-20260704-ge32d26f19115`.
+- Release tag: `sentivent-guidelines-v1.1-20260704-g54aafa445905`.
 - Annotation guidelines version: `v1.1`.
-- Build timestamp: `2026-07-04T13:00:24.095548+00:00`.
-- Builder git commit: `e32d26f19115`.
+- Build timestamp: `2026-07-04T14:52:31.186961+00:00`.
+- Builder git commit: `54aafa445905`.
 - Machine-readable release tag metadata is in `metadata/release_tags.json`.

data/absa_gts_event_sentiment/test.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c8524534aa29cb1b17b3a6ca8e870cb97223046365d5856fdabe8ccbbfb3d80d
-size 212855

data/absa_gts_event_sentiment/train.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:e0ad5a784df791ad3393a2587f6d30058a8e441bbf990b31291c781aa54e046b
-size 1415272

data/absa_gts_sentiment/test.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:73b963fd105c6432bcdb2b0c3ed10e1c080ddf593e3cd8695a89e31dc4a85ba9
-size 106292

data/absa_gts_sentiment/train.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:a8a8375ed5c4474dd661f92be5962aa4e9a4636032c997da7e048ac3b7f19543
-size 836219

data/absa_gts_sentiment/validation.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:4c51a039bcdb5efc98263bc48b1b7f5f6dade52d08b58cf2de6d6feef4782875
-size 120413

data/absa_iabsa_setfit/test.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:aaa3e4944b692bb8de2a0ddd4201a8f1800646adc33f06e183069f8b72be7f30
-size 117279

data/absa_iabsa_setfit/train.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:96115094f8600b84c685139fd9c8a53ed0f4d8539e40ebf4920808e085580f38
-size 780275

data/absa_iabsa_setfit/validation.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b4b0dfbee3a59a8b5ee14c416340771092138a33e2e4e8cc76363d2198f2cb76
-size 137411

data/absa_iabsa_triplet/test.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:ef9d65379f4ad3e0236459cc3e22d6c5b197d599aadc3484ddc34fa36ad9ae23
-size 370524

data/absa_iabsa_triplet/train.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:c3f2de4557cea62ac35609f3606cb8cafe7b4f6ed215c48a50bb673359771bb6
-size 2623211

data/absa_iabsa_triplet/validation.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b0315bdf486ffea002108521aa29e5a0fd188f8e4f3033b873c02869a5da519a
-size 412243

data/absa_implicit_polarity/test.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:14bd09231357fa85d389a51c6a08fc3fa6fc28827ac3bb1f6de345f0207d88f6
-size 129942

data/absa_implicit_polarity/train.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:12caf8b5233827ea1daedda061487be458bb1b2fc674b347a8554148f4f1b8a8
-size 785336

data/absa_implicit_polarity/validation.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:7402b08411a3513e24b5316c1b4c687a747abb05318a2a234dfc593ff6a494f6
-size 117391

data/absa_sentivent_unified_quad/test.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:1082b4d4d7314b9069e94f5ee8a621387413a1b67e40fda356341adbb2844efc
-size 374278

data/absa_sentivent_unified_quad/train.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:711c9a1527bbc169b24676b9b072c889f64c6502e08803bd45f3c0607a6a1ae4
-size 2641294

data/absa_sentivent_unified_quad/validation.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:cd2da841fb725bfdb35c0961ca131d012309d355e017dbf0ffd676670b70e81e
-size 417265

data/documents/test.jsonl

@@ -1,30 +0,0 @@
-{"event_count":32,"event_sentiment_count":0,"id":"aapl14","participant_count":54,"relation_count":14,"sentence_count":15,"sentiment_count":7,"split":"test","text":"iPhone X 's Dangerous Choice Of Market Share Or Profit\nThe decision by Apple to price the iPhone X at one thousand dollars says a lot about Tim Cook 's strategy for the iPhone going forwards .\nAlthough the new iOS-powered handset has been pitched as the future of the smartphone , it 's a future that will see market share falling , even as Apple maintains its profit margin .\nThank to the teardown analysis by IHS Markit reported on by Reuters ' Stephen Nellis we know that the raw cost of the iPhone X is $ 357 .\nComparing that to the list price gives a gross profit margin of sixty-four percent .\nOnce you add in R&D costs , manufacturing , shipping , distribution , and other ancillary costs , the iPhone X should offer Apple the almost traditional forty percent margin on the final handset sale .\nThat 's a lot of margin to enjoy , and there 's an argument that Apple could have sold more units had the price been lower .\nPutting aside stock issues , the iPhone X pricing suggests that Apple will continue to prize profit margins over unit sales and market share .\n2017 has seen modest growth in the overall smartphone market - IHS Markit has reported seven percent year-on-year grow - but Apple 's growth has been lagging the growth of the market , with IHS reporting two percent growth in a similar period .\nTim Cook and his team are facing a similar decision to previous leading manufacturers such as Nokia , Blackberry , Palm , and others .\nGet it wrong and the danger is Apple will slowly follow those companies into the technological history books .\nThe general trend on smartphone pricing has always experienced a downward pressure , and as the price drops more people buy into the system .\nIf the iPhone X had maintained the flagship price of previous years then unit sales and market share numbers could have risen this quarter and throughout 2018 .\nInstead Apple has chosen to maintain its profit margin .\nThat will keep the accountants and shareholders happy , but with no significant year-on-year growth in iPhone sales expected , there 's every chance that 2017 and 2018 will see Apple lose market share to the competition .","token_count":401}
-{"event_count":58,"event_sentiment_count":0,"id":"aapl15","participant_count":137,"relation_count":16,"sentence_count":53,"sentiment_count":24,"split":"test","text":"Here 's how Apple gets to a $ 2 trillion market value\n- Expect strongest growth from services and Asia\nApple is currently the world 's most-valuable company .\nApple 's market capitalization has topped $ 900 billion and seems on track to hit $ 1 trillion by year-end , with expectations for a strong holiday season .\nApple stock is preforming in line with my 2015 prediction , a forecast I repeated again this April , when such a number seemed outlandish .\nSo , what will it take for Apple shares to warrant a $ 2 trillion market cap ?\nListen to Apple CEO Tim Cook and you 'll hear a fair amount of Buck Rogers-sounding futurism : Artificial intelligence that allows for self-driving cars , augmented-reality technology which , demystified , is mostly a more realistic-looking gaming environment for now and more .\nOr you can keep the case for $ 2 trillion simpler : Apple has an already world-class hardware business that sold $ 141.3 billion of iPhones in the year ending this September , while its services business is up to $ 30 billion a year , growing at 23 % .\n`` We expect a lot to happen with two to three years , and it will mostly be in services , '' CFRA Research analyst Angelo Zino says . ''\nThe quickest way from $ 1 trillion to $ 2 trillion is to get a higher stock multiple on top of earnings growth .\nThe longer they grow services at 20 % or better , the better .\nApple will get a better multiple as services become a bigger part of its pie . ''\nThe biggest differences between the prosaic case and the Buck Rogers case lie in how long it might take Apple to double in value .\nBut here 's what to think about when you ask , can Apple really get to $ 2 trillion ?\nFirst , watch services .\nIt 's the most boring part of the case , but the most predictable , Zino says .\nRight now , services are just 13 % of Apple 's revenue , Zino says .\nThis could be a $ 60 billion business within four years , even with some slowing in growth .\nServices could soon produce 20 % of Apple 's revenue and a third of its profit , Zino predicts .\nMorgan Stanley 's Katy Huberty adds that services could even grow 30 % annually if people demand virtual-reality games and other AR-based services .\nServices businesses also command higher multiples , as Zino points out .\nIf we assume a $ 60 billion services business with 33 % profit margins , services alone would generate as much as $ 20 billion in profit .\nApple made $ 64.1 billion before taxes in fiscal 2017 .\nThat could be worth an extra several-hundred billion right there .\nSecond , watch China and Asia .\nIt 's easy to get carried away by China 's potential , Zino cautions .\nApple 's sales to greater China dropped 17 % in fiscal 2016 and another 8 % in 2017 , thanks partly to the financial-market turmoil there .\nApple 's big issue in China is whether its consumers are rich enough to choose iPhones en masse .\nMorgan Stanley 's Huberty argues that Apple 's China recovery , which began earlier this year , will be accelerated by a shift to more-expensive phones like the iPhone X .\nShe says Apple 's strength selling products in China suggests the real problem has been a lack of a compelling new iPhone .\n`` The number of aged iPhones in China due for an upgrade grew 56 % this year , setting up for accelerating upgrades when a new iPhone form factor launches this fall , '' Huberty wrote in August .\n`` What 's more , the switching opportunity from rival phone makers is ripe in China ... .\nThe combination of these factors drives our 66 % bottom-up China iPhone unit growth forecast in FY18 . ''\nBoosting China unit sales by two-thirds would add $ 30 billion in revenue , even if it took longer than a year .\nApple trades at four times sales , so add about $ 120 billion to market cap for that alone .\nThird , watch the tax bill .\nWith $ 252.3 billion in cash held outside of the U.S. , Apple stands to be the biggest beneficiary of Republican proposals to let companies repatriate offshore funds while paying only 10 % as U.S. taxes , well below today 's rate of 35 % .\nThe other thing to watch is the proposal to cut the corporate tax rate to 20 % on U.S. profits .\nApple pays just 25 % now , so a rate cut is n't likely to be a bonanza for Apple , but it would help .\nFourth , watch the car revolution , but do n't count on it .\nNo one knows what Apple will do about cars , Zino says -- but it could be huge .\nNo surprise why : Loup Ventures , co-founded by longtime Apple analyst Gene Munster , says the car market will see more change in the next 20 years than in the last 50 .\nNearly all new cars and light trucks will be electric-powered and mostly or completely self-driving by 2040 , Loup predicts .\nApple probably wo n't make cars , but Zino and Morgan Stanley 's Huberty both see Apple trying to become a major player in the software and services that make autonomous vehicles and driverless-car services practical .\nWill Apple 's ideas outshine Alphabet 's or Uber 's ?\nNo one knows .\nAlso , Loup 's forecast suggests the shift toward autonomous cars will be fairly backloaded , hitting critical mass in the 2030s -- likely too late to drive the climb toward Apple 's $ 2 trillion valuation .\nThe best play is to hold Apple shares if you like the iPhone business , and be patient with the rest .\nThat services will grow steadily is close to a lock .\nIf a tax bill passes , it will help some .\nWith Bank of America Merrill Lynch saying Asian emerging markets face several years of growth before another recession , that rebound in Asia is another decent bet .\nThose positives outweigh the fact that the futuristic Buck Rogers stuff may take a while .\nIf AR and AI do become huge , they 'll push Apple toward $ 3 trillion or more -- even if that now sounds like to infinity and beyond .","token_count":1169}
-{"event_count":41,"event_sentiment_count":0,"id":"aapl16","participant_count":117,"relation_count":11,"sentence_count":34,"sentiment_count":3,"split":"test","text":"Apple 's App Store Generated Over $ 11 Billion In Revenue For the Company Last Year\nApple reported that its App Store generated over $ 26.5 billion in revenue for developers in 2017 , which was up about 30 % year over year .\nThis means that the App Store created approximately $ 11.5 billion in revenue for the company .\nGrowth of Services revenue is one of the major positive points for investors since it is growing faster than the rest of the company with higher margins .\n( Note that I own Apple shares and have sold Put options ) .\nBelow is an analysis of Apple 's App Store revenue that can be gleaned from the past four press releases providing App Store data .\nNote that when I use words such as `` over '' or `` nearly '' they are from the press releases and when I use `` about '' or `` approximately '' they are based on calculations I have done from the numbers in them .\nCalendar 2014\nTotal spending in the App Store was approximately $ 15 billion\nDevelopers received over $ 10 billion in revenue\nApple 's revenue was about $ 4.5 billion\nThe first week of January generated nearly $ 500 million in billings\nJanuary 1st was the largest day\nCalendar 2015\nTotal spending in the App Store was over $ 20 billion\nCumulative developer revenue reached nearly $ 40 billion since 2008 with over one-third in 2015\nDevelopers received about $ 14 billion in revenue , up approximately 35 % year over year\nApple 's revenue was about $ 6 billion , up about 33 % year over year\nThe two weeks ending on January 3 , 2016 , generated $ 1.1 billion in spending\nJanuary 1st saw $ 144 million in total spending\nCalendar 2016\nTotal spending in the App Store was just shy of $ 30 billion\nDevelopers received over $ 20 billion for the year and now over $ 60 billion since 2008\nApple 's revenue was about $ 8.8 billion , up 49 % year over year\nDeveloper 's and Apple 's revenue have both almost doubled in two years\nJanuary 1st had nearly $ 240 million spent that day $ 72 million to Apple\nFor calendar 2016 the App Stores $ 8.8 billion in revenue was about 4 % of the company 's $ 216 billion in total revenue and a third of the $ 25 billion in Services revenue .\nThe almost $ 3 billion increase in App Store revenue made a small dent in Apple 's total revenue drop of about $ 18 billion .\nHowever if it continues to increase at the same rate it does add almost 1.5 percentage points to the company 's total revenue growth for the next few years .\nCalendar 2017\nJanuary 1st had $ 300 million spent that day , up 25 % year over year\n$ 890 million spent over the 7 days from Christmas Eve to New Years Eve\nDevelopers received $ 26.5 billion for the year , up 30 % year over year , and is now over $ 86 billion since 2008\nFor calendar 2017 the App Stores approximate $ 11.4 billion in revenue was almost 5 % of the company 's projected $ 237 billion in total revenue and over a third of the $ 31 billion in forecasted Services revenue .","token_count":592}
-{"event_count":65,"event_sentiment_count":0,"id":"amzn12","participant_count":133,"relation_count":22,"sentence_count":41,"sentiment_count":30,"split":"test","text":"Is Amazon getting into the pharmacy business ? This is what you need to know\nDrugstore chains and pharmacy-benefit managers are likely at risk , Wall Street analysts say\nAmazon.com : books , music , movies , apparel , food -- and prescription medicine ?\nA report earlier this year that the massive online retailer was mulling a pharmaceutical-industry entry has had observers all aflutter .\nAmazon 's nearly $ 14 billion acquisition of Whole Foods , announced in June and closed just over two months later , sparked speculation that the grocery chain might soon house pharmacy locations .\nThough Amazon itself has not confirmed a potential expansion into pharmaceuticals , the possibility has sent stocks of pharmaceutical middlemen south on more than one occasion .\n`` Amazon has shown that they have mastery of the technology and innovative approaches to really transform how the marketplace delivers product , '' said Linda Pissott Reig , who co-chairs the FDA section of the Pittsburgh-based law firm Buchanan , Ingersoll & Rooney PC .\nThe move could also have wider implications for consumers .\nAmazon could use de-identified big data to then see if , for example , `` a patient with depression is also seeking out self-help books , or particular products , '' she said .\n`` Amazon , with a giant marketplace , has a lot of ability to collect information . ''\nWhy the prospect of Amazon 's entry to this new market is important : After transforming such industries as publishing , consumer electronics and cloud services -- and decimating key players along the way -- Amazon has built a reputation for dominance .\nBut part of the wide interest in an Amazon pharmacy is that the highly regulated drug business has never been particularly transparent .\nAt best , U.S. drug pricing is complex ; at worst , it 's dysfunctional .\nThose who are skeptical of Amazon 's potential pharmacy entry have noted Amazon 's difficulty breaking into highly regulated markets , including alcohol distribution in the U.S. Industry leaders have brought up similar points .\nFor one , CVS Health Corp. Chief Executive Larry Merlo told Wall Street analysts in August that his industry was `` highly regulated , so the barriers to entry are high . ''\nCNBC , which first reported on Amazon 's pharmacy plans in May , reported Friday that Amazon would make a decision before Thanksgiving .\nGiven that time table , expect the question to crop up again this earnings season .\nWhat companies could be affected : Though Amazon 's involvement could presumably affect any and all parts of the pharmaceutical supply chain , including drug makers , it 's most threatening to retail pharmacy chains like Walgreens Boots Alliance and Wal-Mart Stores Inc. according to Leerink Partners analyst Ana Gupte .\nThat 's because Amazon is expected to begin getting cash-pay and mail-order customers , eventually moving into the pharmacy-benefits manager and retail pharmacy business , Gupte said .\nCVS , though , might be able to leverage its pharmacy-benefit manager unit to reduce its risk , said Leerink Partners analyst David Larsen .\nAnticipated competition from Amazon could also prompt pharmacies to diversify their offerings , Morgan Stanley analyst Ricky Goldwasser said this summer , including offering lab services like blood tests .\nMiddlemen called pharmacy-benefit managers ( PBMs ) , which work with drug makers to negotiate prices on behalf of health insurers and employers , are another player expected to be affected .\nThey are not a well-known part of the U.S. pharmaceutical system but have come under scrutiny in the last year amid public outrage over drug costs .\nBecause PBMs negotiate discounts from drugs ' total cost , critics charge that they also push prices higher to make their discounts appear more favorable .\nAmazon poses `` both an opportunity and a threat '' for Express Scripts Holding Company , the largest stand-alone PBM , Gupte said .\nExpress Scripts announced on Tuesday that it will buy privately-held medical benefit management company eviCore healthcare for $ 3.6 billion , a move that some saw as a response to the Amazon threat .\nMeanwhile , PBMs that are part of UnitedHealth Group and Humana Inc. will be more shielded , Gupte said .\nDrug distributors such as McKesson Corp. , AmerisourceBergen Corp. and Cardinal Health Inc. will likely have less exposure in the near term , according to Leerink 's Larsen .\nReasons for skepticism : A recent report that Amazon had obtained approval to become a wholesale distributors in several states sent pharmaceutical supply chain stocks down again in late October .\nBut scrutiny of those approvals by the investment firm Jefferies found something important : Correspondence between two state pharmacy boards and Amazon showed that those state licenses could only be used for medical device and supplies distribution .\nThat does n't rule out a pharmaceutical entry , but it does suggest that Amazon is starting with other , potentially easier medical products .\nMizuho analyst Ann Hynes is another skeptic .\nBefore acquiring Whole Foods this year , Amazon had talked about entering the grocery industry for more than a decade , Hynes noted .\nBy contrast , Amazon has n't mentioned pharmacy on its earnings call in the past decade , Hynes said .\nIn addition , even after CNBC reported on the online retailer 's purported plan , the company did not list any pharmacy-related jobs , Hynes said .\nAnd , if the company did decide to move into pharmacy , `` there is no obvious asset to buy , '' Hynes said .\nAmazon would need to either build or buy pharmacy capabilities , Gupte said , and could target a smaller player like specialty pharmacy Avella or acquire a larger entity .\nMany companies might also want to partner with Amazon , Leerink 's Gupte said , including the Blue Cross Blue Shield PBM unit Prime Therapeutics , `` to enhance its drug discounts and mail penetration . ''\nSo might Express Scripts and Walgreens , Larsen said .\nWalgreens shares fell 3.4 % in Oct. 9 trade after the CNBC report , CVS shares declined 3.5 % , Wal-Mart shares rose 2.7 % and shares of Express Scripts plummeted 4.9 % .\nShares of the companies have dropped 13.4 % , 11.7 % , risen 12.4 % and dropped 1.9 % over the last three months , respectively , compared with a 4.5 % rise in the S&P 500 .","token_count":1139}
-{"event_count":28,"event_sentiment_count":0,"id":"amzn13","participant_count":62,"relation_count":7,"sentence_count":17,"sentiment_count":11,"split":"test","text":"Five Reasons Amazon Can Reach $ 1,500\nShares of Amazon ( AMZN ) are higher on Thursday , par for the course as the ecommerce giant has trounced many traditional retailers , and Cowen & Co. thinks they will keep rising still .\nAnalyst John Blackledge and his team reiterated an Outperform rating on the stock today and raised their price target from $ 1,225 to $ 1,500 , calling it their best idea for 2018 .\nAlthough Amazon has already been on a tear this year , rising nearly 55 % since the start of 2017 , they think that the stock will be able to keep climbing , thanks to catalysts across its business , from the cloud to ecommerce .\nThey city five main drivers in their note :\n`` (i) Further B2C eCommerce market share gains in large retail verticals like Apparel , Consumables , & Food & Bev Grocery , key drivers of the eCommerce business near and intermediate term ;\n(ii) Emerging eCommerce verticals like B2B , where AMZN recently introduced Prime for Business for small businesses to enterprise customers , signaling its emergent position in the estimated ~ $ 1.1 TN US , UK and German B2B eCommerce markets .\nAMZN 's current US B2B product mix is diverse , offering Medical Supplies to Precision Parts to Consumables / Groceries with ability to service the opportunity within existing fulfillment infrastructure ;\n(iii) Significant Int ' l opportunity in existing and newer markets like India $ 1.1 TN retail sales , ~ 1 % eCom penetration , Mexico $ 426BN retail sales , ~ 2 % eCom penetration , and Australia $ 253BN retail sales , 5 % eCom penetration ;\n(iv) AWS is the market leader in Public Cloud with an $ 18BN revenue run-rate and ~ 25 % operating margins .\nAWS should enjoy years of secular tailwinds driving revenue CAGR of ~ 26 % ' 18 - ` 23 as workloads migrate to the Cloud ;\n(v) AMZN Advertising , while still nascent , offers a dual value prop , a revenue growth opportunity driven by great user targeting given vast 1P user data , and b significant margin opportunity . ''\nThey write that despite Amazon 's rally , the market is still underestimating the company , as it 's one of the few in the world that can add $ 100 billion or more in addressable opportunities by expanding into new markets domestically and internationally , while building its share in areas where it already competes .\nThey also see Amazon Prime as a long-term growth driver .\n( Earlier this week , Instinet said it was bullish about Amazon 's apparel business as well . )\nShares of Amazon are up 0.5 % to $ 1157.87 this afternoon .\nThe SPDR S&P Retail ETF XRT is higher as well .","token_count":501}
-{"event_count":18,"event_sentiment_count":0,"id":"amzn14","participant_count":46,"relation_count":4,"sentence_count":13,"sentiment_count":4,"split":"test","text":"Amazon sold more Echo Dots than any other item over the holidays\nAlexa was the hottest gift on Amazon this holiday season .\nThe Echo Dot , which stars Amazon 's cheerful virtual assistant , was the best-selling item on all of Amazon during the 2017 winter shopping period , the online-retail giant said in a release this week .\nMillions of Echo Dot devices were sold worldwide -- more than any other item by any manufacturer in any category on the site .\nTens of millions of Alexa-enabled devices -- which includes the Echo Dot -- were sold on Amazon overall , the company said .\nSome models , like the new Echo Spot , the Echo Dot , and Echo Buttons accessories sold out , and are now only available for post-holiday pre-order .\nThe Fire TV Stick with Alexa Voice Remote , a streaming device , was also a popular buy .\nIf we are conservative and estimate that Amazon sold 20 million of its best seller , the Echo Dot , at the holiday sale price of $ 29.99 , that would equal around $ 600 million in sales , research firm Nomura 's Instinet said in a report .\nThe number may be even higher give that other Echo devices , like the Show and Spot , cost upwards of $ 200 and $ 100 , respectively .\nAmazon 's markdowns and more affordable options may have helped introduce Alexa to more homes this holiday season .\nThe Echo Dot , which usually sells for $ 50 and is the cheapest Echo device on the market , was on sale for $ 29.99 during the holidays -- a 40 % discount .\nThe regular echo was marked down to $ 79.99 .\nAnd the Echo Spot was introduced this year as a more affordable and compact version of the Echo Show , which has camera and video capabilities .","token_count":331}
-{"event_count":32,"event_sentiment_count":3,"id":"ba14","participant_count":70,"relation_count":13,"sentence_count":30,"sentiment_count":30,"split":"test","text":"Boeing 's Low-Altitude Bid\nBoeing Co. 's purported bid for Embraer SA looks low .\nBoeing has been aggressively pursuing a takeover of the Brazilian regional jet maker , a remarkable shift for a company that dubbed a similar combination struck by Airbus SE and Bombardier Inc. in October `` a questionable deal '' that would n't force it to change its strategic path .\nMany details are still being sorted , the most important being whether the Brazilian government , which has veto power , will support an acquisition .\nBut at least one thing has reportedly been worked out : the price .\nAccording to the Wall Street Journal , Boeing and Embraer have informally agreed to a deal at ... $ 28 a share .\nThat 's hardly the `` relatively large '' premium that previous reports had indicated was coming .\nA bid of $ 28 would be 40 percent higher than where Embraer was trading before news emerged of Boeing 's interest .\nThat 's not insignificant , but also not atypical in M&A , especially lately .\nIt 's also worth noting that Embraer 's stock had taken a hit this fall as the company faced the prospect of heightened competition from the Bombardier-Airbus combination , a problem it could better handle as part of Boeing .\nFalling Short\nInvestors were clearly expecting more .\nThe American depositary receipts had climbed to about $ 27 heading into Friday , a level that 's untenable if the ultimate price is just $ 1 higher because the deal has n't even officially been inked yet and faces significant regulatory hurdles .\nThus , news of the price sent Embraer 's ADRs plummeting as much as 7 percent .\nShareholders had every right to be more optimistic .\nEmbraer is a crown jewel for Brazil .\nWhile Boeing is still negotiating with the government over what would happen with Embraer 's defense business and how much say politicians would have going forward , a hefty offer was assumed to be part of its argument .\nThe regional-jet hole in Boeing 's lineup has been laid bare by its trade dispute with Bombardier over claims the Canadian company used low prices to unfairly compete for a Delta Air Lines Inc. order for which Boeing was n't even a contender .\nIf Boeing is serious about catching up to the new Airbus-Bombardier union , it does n't have many other options that match the scale of Embraer .\nAnd yet Boeing 's offer values Embraer 's ADRs at a discount to the median multiple paid in major aerospace and defense deals over the past decade , according to data compiled by Bloomberg .\nTakeovers targets of more than $ 1 billion have typically commanded about 13 times Ebitda .\nRelatively ( Not ) Large\nBoeing 's reported bid for Embraer is just a modest bump up from where the company had already climbed on news of its interest .\nThe valuation wo n't take your breath away .\nNot all of the acquisitions in this industry are perfect comparisons ; Embraer has no hope of fetching the kind of pricey valuation that higher-margin avionics supplier Rockwell Collins Inc. did from United Technologies Corp. last year , for example .\nBut Lockheed Martin Corp. paid 13 times Ebitda to acquire helicopter maker Sikorsky from United Technologies in 2015 , or a little over 10 times if you adjust for the benefits of tax maneuver .\nSomething in that ranges seems more reasonable for Embraer .\nBoeing still has work to do to secure an Embraer acquisition .\nPerhaps that should include re-evaluating the price .\nThe Brazilian government is n't the only stakeholder it has to please .","token_count":656}
-{"event_count":23,"event_sentiment_count":5,"id":"ba15","participant_count":33,"relation_count":13,"sentence_count":13,"sentiment_count":9,"split":"test","text":"Should Boeing Buy GE Aviation ?\nBoeing ( BA ) was the Dow Jones Industrial Average 's best-performing stock in 2017 , and it 's gained nearly 20 % so far in 2018 .\nAfter such a monster runup , investors may be asking how the airplane maker can keep the rally going .\nMelius Research 's Scott Davis thinks that mergers and acquisitions may be the answer .\nHe reiterated his Buy-Accumulate rating on Boeing and raised his price target to $ 500 Wednesday , writing that the company is still in the early innings of what he believes is a `` once-in-50-year transformation '' on a wide scale that will make it a higher-return business with lower risk .\nWhile Boeing is making major changes in terms of its businesses , Davis thinks that some strategic M&A could complement these efforts .\nWhile Boeing has been rumored to be on the hunt for some time , he believes that the company may finally be about to make a move .\nAnd what 's the old saying -- go big or go home ?\nHe sees the former happening , and what better big move , he asks , than buying General Electric 's ( GE ) GE Aviation unit ?\nHe writes that it may not be the easiest deal to swallow , but Boeing can afford it , and if it could overcome regulatory and antitrust objections , it could be `` groundbreaking . ''\nGE , the worst-performing stock in the Dow last year , certainly could be persuaded to part with GE Aviation , and Davis believes that the Trump administration is `` Boeing-friendly '' and that the synergies of such a deal `` would be gigantic . ''\nBoeing is up 0.2 % in early trading .\nThe SPDR S&P Aerospace & Defense ETF ( XAR ) is flat .","token_count":338}
-{"event_count":13,"event_sentiment_count":0,"id":"ba16","participant_count":31,"relation_count":5,"sentence_count":7,"sentiment_count":0,"split":"test","text":"Boeing 's stock contributes about 10 % of the Dow 's 1,030-point tumble\nShares of Boeing Co. were exacting the biggest toll on the Dow Jones Industrial Average on Thursday .\nThe plane manufacturer , which has been the most influential component of the price-weighted Dow was contributing almost 115 points of the blue-chip average 's drop .\nBoeing 's shares were off 4.8 % , or 16.75 .\nA $ 1 move in any one of the Dow 's 30 components equates to a $ 6.83-point slide in the benchmark .\nShares of Caterpillar Inc. Goldman Sachs Group Inc. and Home Depot Inc. were also pushing the Dow decidedly lower by a combined 140 points .\nMeanwhile , the S&P 500 index ended 3.8 % at 2,582 , while the Nasdaq Composite Index closed 3.9 % lower at 6,777 .","token_count":156}
-{"event_count":79,"event_sentiment_count":3,"id":"bac04","participant_count":116,"relation_count":44,"sentence_count":42,"sentiment_count":17,"split":"test","text":"Bank of America Earnings Hurt by Tax-Related Charge\nBank of America has made a concerted effort to cut costs and focus on safer businesses such as lending to consumers with good credit .\nAbove , customers use ATMs at a bank in New York .\nBank of America has made a concerted effort to cut costs and focus on safer businesses such as lending to consumers with good credit .\nBank of America Corp. said a charge from the new U.S. tax law caused quarterly profit to fall by 48 % , even as the bank ended 2017 with its crisis-era issues firmly in the past .\nThe Charlotte , N.C.-based bank on Wednesday reported fourth-quarter profit of $ 2.37 billion , or 20 cents a share .\nThat was down from $ 4.54 billion a year earlier .\nWithout the $ 2.9 billion tax charge , however , the bank 's profit would have risen to $ 5.3 billion , or 47 cents a share .\nAnalysts polled by Thomson Reuters had expected earnings of 44 cents a share on an adjusted basis .\nFor the year , the bank posted a $ 21.1 billion profit , excluding the tax adjustment .\nThat roughly matched the bank 's all-time profit record from 2006 , when it was a simpler consumer-focused firm that had n't yet bought ailing mortgage lender Countrywide Financial Corp. or investment bank Merrill Lynch & Co.\nThe bank 's executives said Wednesday that the 2018 outlook is bright .\nThat is because the one-time hit from the tax bill is expected to be outweighed by the law 's longer-term benefits .\nThe fourth-quarter charge was largely due to the bank writing down its deferred tax assets .\nThese credits to offset future tax bills were created by past losses , in many cases huge ones racked up during the financial crisis .\nThe assets generally lose value when tax rates fall .\nThe bank expects its effective tax rate for 2018 to be 20 % , down from an expected 29 % before the new tax law , Chief Financial Officer Paul Donofrio said on a call with reporters Wednesday morning .\nChief Executive Brian Moynihan said he expected most of the bank 's benefit from the tax cut would go to shareholders , who should expect higher capital returns in the form of dividends or share buybacks .\nSome of it will be spent on investments .\nMr. Moynihan also reiterated that he expected tax changes would eventually lead to more loan growth .\nThe benefits from the tax bill are expected to accelerate the progress Bank of America has made rebuilding its business after the financial crisis .\nAs recently as 2014 , the lender 's results were dogged by tens of billions of dollars in penalties over financial-crisis era issues .\nSince then , the company 's legal problems have eased , as Mr. Moynihan has made a concerted effort to cut costs and focus on safer businesses such as lending to consumers with good credit .\nThe boost from the tax bill will expedite the bank 's timeline for meeting long-held performance goals of a 1 % return on assets and a 12 % return on tangible common equity , Mr. Moynihan said .\nExcluding the effect of the tax bill , those metrics stood at 0.9 % and 10.9 % in the fourth quarter .\nThe bank has recently had the help of rising interest rates , which are boosting profits .\nBank of America 's net-interest income was $ 11.46 billion , up about 11 % from a year earlier .\nIt paid slightly higher rates to depositors in the quarter , though holders of regular savings accounts are still getting almost nothing in interest .\nThe rate the bank paid on U.S. interest-bearing deposits was 0.27 % , compared with 0.24 % in the prior quarter .\nIt also cut expenses to $ 54.74 billion on the year , coming closer to a $ 53 billion target the bank has set for 2018 .\nRevenue for the year rose 4 % to $ 87.4 billion .\nThere were still some hiccups in the quarter .\nOne was a $ 292 million charge related to `` a single-name non-U.S. commercial '' client in the fourth quarter .\nA person familiar with the matter said this came from the bank 's lending activity involving troubled firm Steinhoff International Holdings NV .\nJPMorgan Chase & Co. and Citigroup Inc. both took similar charges for loans involving the retailer , which is battling burgeoning financial problems after disclosing possible accounting irregularities .\nTrading revenue was another weak spot , as it has been for other large U.S. banks including Goldman Sachs Group Inc. .\nExcluding an accounting adjustment , Bank of America 's trading revenue fell about 9 % to $ 2.66 billion from $ 2.91 billion in last year 's fourth quarter , though that was less than the bank initially predicted .\nThe lender 's improving fortunes recently helped lift its stock above $ 30 a share for the first time since 2008 .\nShares are up more than 80 % since the 2016 presidential election , when hopes for deregulation , tax cuts and rate increases sent stocks in the sector higher .\nShareholders greeted Bank of America 's latest results with caution , sending the bank 's shares 0.8 % lower in morning trading .\nThe bank had to issue so many new shares to deal with its crisis-era problems that per-share profits remain far below pre-2008 levels .\nAnd Bank of America still trades at a lower valuation than some competitors like JPMorgan Chase .","token_count":992}
-{"event_count":12,"event_sentiment_count":0,"id":"bac05","participant_count":22,"relation_count":7,"sentence_count":10,"sentiment_count":3,"split":"test","text":"BofA Includes Bitcoin Trust in Broader Ban on Investments\nBank of America Corp. 's Merrill Lynch told employees last month not to offer clients Grayscale 's Bitcoin Investment Trust , one of the few financial instruments directly holding the digital coin , as the brokerage broadly eschews the virtual currency .\nThe firm -- already known to be refraining from offering bitcoin futures contracts -- told financial advisers not to pitch the fund or execute new client requests to buy into it , according to an excerpt of a Dec. 8 internal memo .\nThe bank is willing to maintain existing positions in brokerage accounts , but not in fee-based advisory accounts .\nThe Wall Street Journal reported the decision earlier on Wednesday .\nWall Street firms have broken ranks in recent weeks over whether to facilitate customers ' bets on bitcoin amid widespread concern that it 's a bubble or that the cryptocurrency 's price is potentially susceptible to manipulation .\nBank of America is among brokerages that have held off on offering the futures introduced last month by Cboe Global Markets Inc. and CME Group Inc. .\nThe bank issued its policy on the trust days before the initial futures contracts began trading on Dec. 10 .\n`` The decision to close GBTC to new purchases is driven by concerns pertaining to suitability and eligibility standards of this product , '' executives wrote in the memo .\nThe bank does n't offer retail clients any other bitcoin-linked products , according to a person with knowledge of the matter who asked not to be identified discussing company offerings .","token_count":283}
-{"event_count":50,"event_sentiment_count":0,"id":"bac06","participant_count":89,"relation_count":28,"sentence_count":32,"sentiment_count":20,"split":"test","text":"Bank of America Hires Law Firm to Help Probe $ 292 Million Loss\nBank of America has hired Davis Polk & Wardwell to look into a $ 292 million charge tied to a soured loan to the parent company of Mattress Firm and Sleepy 's .\nBank of America Corp. has brought in an outside law firm to help examine a soured lending arrangement that led to a $ 292 million charge in last year 's fourth quarter , according to people familiar with the matter .\nThe losses sprang from financing involving troubled South African firm Steinhoff International Holdings SNH 8.66 % NV , the people said .\nBank of America is trying to figure out if the losses , which surfaced at many big global banks , could somehow have been avoided , the people said .\nThe inquiry is being conducted by law firm Davis Polk & Wardwell LLP and began late last year , the people said .\nThe law firm recently has interviewed Bank of America staff , some of the people added .\nBank of America 's board is being kept abreast of the inquiry , which is ongoing , some of the people said .\n`` One of the reasons we have record-low credit losses is because we take the time to analyze what happened when things do n't go as planned and learn from it .\nIt 's the responsible thing for a financial institution to do , '' a bank spokesman said .\nBank of America Chief Executive Brian Moynihan alluded to the inquiry on the bank 's earnings call in January .\n`` It is always a wake-up call that some things do n't turn out well , and we got to go back and what are lessons learned and what did we do right or wrong on that and how we avoid that in the future , '' Mr. Moynihan said of the loss .\n`` We were n't happy with it , from the top of the house through to the actual people who were involved in it . ''\nBank of America was one of a number of global lenders that held pieces of a € 1.6 billion ( $ 2.0 billion ) loan to Christo Wiese , then chairman of the South African retailer .\nMr. Wiese 's shares in the company served as collateral for the financing .\nThose shares collapsed in value last fall after the retailer , which owns Mattress Firm and the Sleepy 's brand , disclosed accounting irregularities .\nThe speed of the resulting share-price decline -- the stock was worth 120 % of the loan on Dec. 5 and less than 24 % two days later -- left banks few good options .\nIn addition to Bank of America , Citigroup Inc. , JPMorgan Chase & Co. and Goldman Sachs Group disclosed losses to what was typically described as a single-name client in the fourth quarter .\nThis client , people familiar with the matter said , was Steinhoff .\nThe four banks booked a combined hit to fourth-quarter earnings of more than $ 1 billion .\nBank of America 's $ 292 million loss was the second-largest charge for the quarter and was spread across its global banking and markets divisions .\nThat drove the bulk of loan-loss provisions in both units in the fourth quarter , a sign of both how large the Steinhoff loss was -- and how healthy the rest of the units ' portfolios were by contrast .\nCitigroup had the highest charge , described as up to $ 370 million .\nThe losses hit banks so quickly and broadly that many analysts have brushed off the charges as a one-time occurrence that is n't indicative of broader underwriting problems .\nStill , Bank of America is looking to see how the exposure to Steinhoff came to be and how the bank can avoid similar problems in the future , the people familiar with the matter said .\nCitigroup , Goldman Sachs , HSBC and Nomura put together the original loan to Mr. Wiese .\nThose banks sold off large chunks of their exposure to several other banks , including Bank of America .\nThe loan was made to a vehicle controlled by Mr. Wiese in September 2016 to help him fund a Steinhoff capital-raising needed to pay for one of its many acquisitions .\nMr. Wiese 's vehicles pledged 628 million shares -- at that point worth $ 3.175 billion -- as collateral against the loan .\nIn essence , Mr. Wiese borrowed against his Steinhoff shares to buy more Steinhoff shares .\nSuch stock-based lending had been booming across the banking industry due to what was until recent days a long , low-volatility stock rally and the loans ' light capital requirements .\nBank of America , which experienced major losses and fines in connection with the financial crisis , has become much more conservative in recent years across its businesses .","token_count":869}
-{"event_count":32,"event_sentiment_count":0,"id":"cvx04","participant_count":83,"relation_count":7,"sentence_count":24,"sentiment_count":9,"split":"test","text":"Strong Crude Oil No Help for Chevron , Exxon Mobil\nNymex crude oil ended last week at $ 65.45 per barrel , up 8.9 % year to date , outperforming the gain of 3.2 % for the Dow Jones Industrial Average .\nChevron Corporation ( CVX ) and Exxon Mobil Corporation ( XOM ) are two '' Dogs of the Dow '' for 2018 , and they ended last week with Chevron shares down 5.3 % and Exxon Mobil shares up just 1.1 % year to date .\nChevron and Exxon Mobil both reported quarterly earnings on Friday , and both companies missed analysts ' expectations .\nAs `` Dogs of the Dow , '' what 's more important than earnings is the strategy to buy weakness on brand-name dividend stocks .\nChevron has a dividend yield of 3.78 % , and Exxon Mobil has a dividend yield of 3.71 % , which make these stocks too cheap to ignore based on dividends .\nHowever , the oil giants are not cheap looking at their P/E ratios .\nChevron 's P/E is 34.52 , and Exxon Mobil 's is 27.60 .\nThe P/E for the Dow 30 is 26.85 .\nThe Weekly Chart for Crude Oil\nThe weekly chart for crude oil is positive but overbought , with oil above its five-week modified moving average of $ 62.67 .\nOil is also above its 200-week simple moving average at $ 56.28 and has been above this `` reversion to the mean '' since the week of Dec. 29 , when the average was $ 57.34 .\nThe 12 x 3 x 3 weekly slow stochastic reading ended last week at 91.97 , above the overbought threshold of 80.00 and above 90.00 as an `` inflating parabolic bubble . ''\nGiven this chart and analysis , my strategy is to buy oil on weakness to my monthly value level of $ 61.69 and to reduce holdings on strength to my weekly risky level of $ 67.60 .\nI show annual and quarterly pivots of $ 63.81 and $ 64.53 , respectively .\nThe Weekly Chart for Chevron\nThe weekly chart for Chevron is negative , with the stock below its five-week modified moving average of $ 123.50 and above its 200-week simple moving average of $ 106.95 , which is the `` reversion to the mean , '' last tested during the week of Aug. 25 , when the average was $ 107.13 .\nThe 12 x 3 x 3 weekly slow stochastic reading is projected to slide to 68.49 this week , falling below the overbought threshold of 80.00 .\nGiven this chart and analysis , I recommend buying Chevron shares on weakness to my semiannual value level of $ 98.73 and reducing holdings on strength to my monthly risky level of $ 129.72 .\nThe weekly chart for Exxon Mobil\nThe weekly chart for Exxon Mobil is projected to be negative at the end of this week if the stock closes the week below its five-week modified moving average of $ 84.90 and below its 200-week simple moving average of $ 86.35 , which is the `` reversion to the mean . ''\nThe 12 x 3 x 3 weekly slow stochastic reading is projected to end this week at 76.40 , falling below the overbought threshold of 80.00 .\nGiven this chart and analysis , my strategy is to buy Exxon Mobil shares on weakness to my semiannual value level of $ 73.53 and to reduce holdings on strength to my quarterly and annual risky levels of $ 92.47 and $ 103.71 , respectively .\nMy monthly pivot is $ 84.82 .","token_count":643}
-{"event_count":30,"event_sentiment_count":0,"id":"cvx05","participant_count":92,"relation_count":21,"sentence_count":30,"sentiment_count":10,"split":"test","text":"Chevron 's 10-K Puts the Permian on a Pedestal\nShale helped the oil major replace reserves at a healthy pace last year .\nChevron Corp. is not shy about touting its 1.7 million acres of real estate in the prolific Permian basin , the center of the U.S. tight-oil boom .\nAnd its annual filing , which dropped late Thursday , showed just how big the basin figures in one important respect : reserves .\nReplacing reserves is an obsession for oil majors ; it 's hard to be an oil major for long if you do n't replenish what you pump .\nAccording to its 10-K , Chevron replaced 161 percent of its production organically -- before factoring in purchases and disposals -- a healthy figure after several pretty mediocre years :\nBooked Up\nChevron 's additions to proved reserves took a quantum leap in 2017 , relative to production\nFor that , Chevron can thank the U.S. and its Permian position in particular .\nThe company 's home operations added almost 6 cubic feet of natural gas to proved reserves for every one produced and almost four barrels of oil for every one pumped out .\nHome Advantage\nChevron 's U.S. operations powered the big jump in the company 's reserve-replacement ratio in 2017 Note : Organic reserve-replacement ratio .\nBy extension , Chevron can also thank OPEC , the Russians and a handful of other countries cutting supply and thereby pushing average benchmark U.S. oil prices up by about $ 7.40 a barrel last year .\nRoughly a third of the additions last year were revisions to existing reserves , where oil and gas prices play a significant role ( because they can make reserves economically viable or unviable at any given year-end ) .\nAnd a better price environment encourages oil and gas companies to get back to work ; Chevron drilled or participated in 310 wells in the Permian basin last year versus just 201 in 2016 .\nConsequently , it added almost as many barrels of oil equivalent to its U.S. proved reserves via extensions and discoveries in 2017 as it did in the prior three years combined .\nThis was important , especially , in the crucial half of the reserves equation : oil .\nChevron 's additions to lower-value natural gas reserves were split fairly evenly between the U.S. and international areas .\nBut in liquids , it was definitely a case of America first :\nTexas Tea\nChevron 's replacement of oil production in 2017 was dominated by the U.S. Note : Additions are organic reserve replacement , excluding purchases and disposals .\nThis shift toward home is n't confined to Chevron .\nExxon Mobil Corp. has n't filed its 10-K yet , so details on reserves are scarce .\nStill , we know from its announcement earlier this month that about 800 million -- or almost a third of the 2.7 billion barrels of oil equivalent added to its proved reserves in 2017 -- came from its unconventional operations in the U.S. , mainly in the Permian basin .\nChevron , however , has an advantage here .\nIt has been established in the Permian basin for a long time and consequently pays low or no royalties on its acreage there .\nExxon , meanwhile , has been buying its way into the oil world 's equivalent of Manhattan real estate , in part to help fill a project pipeline looking somewhat thin due to sanctions on Russia .\nBarclays estimates Exxon 's average royalty rate in the Permian at about 20 percent , versus less than 10 percent for Chevron .\nBoth companies have yet to fully demonstrate they can make the capital-intensive business of shale development work for a supermajor 's needs read : dividends .\nChevron , however , would appear to have the edge in terms of taking those proved reserves and proving itself .","token_count":690}
-{"event_count":50,"event_sentiment_count":0,"id":"cvx06","participant_count":76,"relation_count":19,"sentence_count":29,"sentiment_count":2,"split":"test","text":"Chevron 's Debt Fell in 4Q17 : What to Expect in 2018\nIs Chevron Set to Grow in 2018 ?\nChevron 's debt position compared to peers\nIn this series , we 'll look at Chevron 's CVX current financial position and growth prospects .\nWe 'll begin with debt and cash flow analysis .\nThen we 'll look at the capex , segmental dynamics , and upstream growth path .\nLet 's look at CVX 's debt position first .\nChevron 's net debt-to-adjusted EBITDA ratio stood at 1.4 x in 4Q17 , below the average industry ratio of 1.6 x .\nThe industry average takes into account 13 integrated energy companies worldwide .\nAnother parameter to compare debt is the total debt-to-total capital ratio .\nIn 4Q17 , CVX 's total debt-to-total capital ratio stood at 21 % , again below the industry average of 36 % .\nChevron 's net debt fell from $ 39.1 billion in 4Q16 to $ 33.9 billion in 4Q17 , which represents a 13 % YoY fall .\nRoyal Dutch Shell ( RDS.A ) saw an 11 % YoY fall in net debt in 4Q17 .\nExxonMobil 's ( XOM ) net debt levels remained stable in the period .\nHowever , BP ( BP ) saw a rise in its net debt levels of 6 % YoY in 4Q17 .\nAnalyzing Chevron 's debt trend\nCVX 's net debt-to-adjusted EBITDA ratio fell from 2.6 x in 4Q16 to 1.3 x in 4Q17 .\nBefore examining the trend , let 's see why CVX 's net debt fell .\nThe fall in CVX 's net debt was because of a steeper decline in total debt than the fall in cash and equivalents .\nTotal debt fell by $ 7.4 billion from 4Q16 to $ 38.7 billion in 4Q17 .\nThe fall in total debt was due to an increase in upstream earnings resulting from higher oil prices .\nCash and equivalents fell by $ 2.2 billion from 4Q16 to $ 4.8 billion in 4Q17 .\nCVX 's adjusted EBITDA rose in 4Q16 -- 4Q17 because of higher earnings in its upstream segment .\nThe rise in its adjusted EBITDA , along with a decrease in its net debt from 4Q16 to 4Q17 , led to a decline in Chevron 's net debt-to-adjusted EBITDA multiple .\nWhat does Chevron 's debt analysis imply ?\nChevron stands below the peer average for both the debt ratios we discussed above -- a favorable scenario .\nPlus , CVX 's total debt-to-capital ratio is the second-lowest among its peers , placing it in a comfortable leverage position that provides Chevron financial strength and flexibility to handle difficult times .\nPlus , CVX 's net debt-to-EBITDA ratio has declined consistently since 4Q16 due to the rise in oil prices , which led to a surge in upstream earnings .\nGoing forward , if oil prices rise and Chevron implements its financial strategy of reducing costs , optimizing capex , and divesting non-core assets , then Chevron could see a further fall in its net debt-to-EBITDA ratio .","token_count":544}
-{"event_count":13,"event_sentiment_count":0,"id":"duk05","participant_count":35,"relation_count":9,"sentence_count":12,"sentiment_count":6,"split":"test","text":"Duke Energy Says Some Customers May Be Affected by Data Breach\nDuke Energy Corp. said Tuesday some of its customers may have been affected by a data breach revealed by PayPal Holdings Inc. last week .\nOn Friday , Tio Networks , which PayPal bought in July , said it had uncovered evidence of unauthorized access to its network that could have compromised personally identifiable information for roughly 1.6 million users .\nAmong customer information possibly exposed , according to a PayPal spokesman , were names , addresses , bank account details , Social Security numbers and details of login information used to pay bills .\nDuke said some 374,000 Duke Energy Carolinas customers who paid a bill by check or cash at one of its 550 authorized walk-in payment processing centers from 2008 through this year may have been affected by the breach .\nCustomers who paid by credit card were n't affected , Duke said .\nLast month , the payments firm suspended operations , pending a security review , of its TIO Networks unit .\nThat company makes digital bill-payment tools for utilities and other firms and also operates a network of kiosks in physical retail stores .\nLesley Quick , Duke Energy 's vice president of revenue services , said Tuesday the company has been in daily contact with the vendor since it `` abruptly and unexpectedly disabled their network on Nov. 10 for suspected ' security vulnerabilities .\n''' The company also said the breach could affect other companies using TIO Networks to process payments .\nShares in Duke were down 0.8 % to $ 87.90 near the end of regular trading Tuesday .\nPayPal was up 0.4 % at $ 71.24 .","token_count":295}
-{"event_count":8,"event_sentiment_count":0,"id":"duk06","participant_count":16,"relation_count":0,"sentence_count":10,"sentiment_count":3,"split":"test","text":"Like Many of Its Peers , DUK Is Trading in the Oversold Zone\nAccording to technical analysts , a stock is considered to be trading in the oversold zone when its RSI falls below 30 .\nIt 's considered to be trading in the overbought zone when its RSI rises above 70 .\nExtreme RSI levels may indicate an imminent reversal in the direction of a stock .\nNote that the stocks of most utility giants , such as NextEra Energy NEE , Southern Company SO , and Dominion Energy D , are currently trading in the oversold zone .\nShort interest in Duke Energy stock rose a massive 20 % on December 15 , 2017 .\nAccording to a recent report , total shorted shares in DUK were 7.3 million on November 30 , 2017 , and this number rose to 8.5 million as of December 15 , 2017 .\nAn increase in short interest might mean that more investors are expecting a stock to fall from its current levels .\nShort interest is a measure that shows investor nervousness .\nShares that are sold short and not yet covered back are represented as short interest .","token_count":199}
-{"event_count":21,"event_sentiment_count":0,"id":"duk07","participant_count":41,"relation_count":13,"sentence_count":21,"sentiment_count":4,"split":"test","text":"Duke Energy Stock Is at Its Most Oversold Level in 5 Years\nHas Duke Energy Stock Turned Cheap after Its Recent Fall ?\nRelative strength index\nBroader utilities ( XLU ) ( IDU ) have been falling almost every day so far in 2018 .\nThe second-largest utility by market cap , Duke Energy ( DUK ) , has been on a similar downtrend in this period .\nIn the last month , DUK has fallen nearly 12 % .\nDuke Energy stock is currently trading deep in the oversold zone with an RSI relative strength index of 5 .\nDUK has n't seen such oversold levels in the last five years .\nAn RSI at such a low level could trigger a reversal in the stock 's direction .\nMoving averages\nOn January 18 , 2018 , Duke Energy stock was trading 10 % below its 50-day and 200-day moving averages .\nThe huge discounts to both the stock 's key moving averages suggest underlying weakness in the stock .\nThe price level of ~ $ 85.5 could act as a resistance for Duke stock going forward .\nThe stock may continue to trade weakly until it stays below its moving averages .\nIt 's currently trading at $ 77.3 .\nShort interest\nShort interest in Duke Energy stock fell 6 % on December 29 , 2017 , according to a recent report .\nThe total number of shorted shares in DUK was 8.7 million on December 15 , and shorted shares fell to 8.1 million by December 29 , 2017 .\nThe fall in the stock 's short interest could mean that fewer investors are expecting it to fall from its current level .\nShort interest represents the total number of a company 's shares that have been sold short and have n't yet been covered back .\nThe metric tracks investor anxiety .","token_count":325}
-{"event_count":30,"event_sentiment_count":0,"id":"f13","participant_count":64,"relation_count":15,"sentence_count":39,"sentiment_count":13,"split":"test","text":"Ford Rolls Out a Hot Rod SUV as Drivers Abandon Performance Cars\nEdge ST aimed at buyers who once sought high-horsepower sedans\nAutomaker sees ` a pretty healthy business ' in souped-up SUVs\nFord Motor Co. has long souped up its sedans to satisfy customers looking for high performance .\nCar demand is waning now but the need for speed is n't , so the automaker is rolling out a fresh hot rod that 's a sign of the times : its first race-ready SUV .\nThe ST edition of the Edge sport utility vehicle that Ford introduces Thursday in advance of the Detroit auto show is powered by a 335-horsepower twin-turbo V-6 engine , engineered on the track and styled to look `` sinister , '' said its designer Steve Gilmore .\n`` We shredded a lot of race tracks with this vehicle , '' Jonathan Crocker , the Edge 's chief engineer , told reporters ahead of the reveal .\nIf the idea of doing hot laps in the family hauler seems slightly absurd , welcome to the latest twist in the SUV craze .\nAs U.S. car buyers continue their mass migration into utility vehicles , some are bringing along their preference for performance once sated by Taurus SHO sport sedans and Focus ST hot hatches .\nNow Ford , better known for muscling up its Mustang pony car , sees a new way to sell more high-profit SUVs .\nMore SUVs Ahead\n`` I would n't describe it as a niche , '' Raj Nair , head of Ford 's North American operations , said in an interview .\n`` We see it as big enough to pursue some real engineering and some real unique parts .\nFor that to pay off , it needs to be a pretty healthy business . ''\nThe Edge will be just the first in a series of hot rod SUVs Ford has coming , Nair said .\nAnalysts speculate the compact Escape and subcompact EcoSport will be next in line for the ST upgrade .\n`` People are running away from cars and they 're running toward trucks and SUVs , '' said Karl Brauer , executive publisher of Cox Automotive Inc. brands including car-shopping website Autotrader and researcher Kelley Blue Book .\n`` Why not have the performance elements of cars migrate to SUVs , too ? ''\nFiat Chrysler Automobiles NV has already blazed the trail with SRT and R/T editions of its Dodge Durango large SUV and a Jeep Grand Cherokee that comes in a high-performance SRT package or a Trackhawk edition with a monstrous Hemi V-8 engine .\nAbout 18 percent of Durango SUVs registered last year through October were SRT and R/T models , according to IHS Markit data .\nFemale Following\nAn understated mid-sized people mover , the Edge may seem an unlikely place for Ford to start spicing up its SUV line .\nIt attracts the largest female following among Ford 's SUVs , with nearly half bought by women , the company said .\nIts primary competitor has been the smoothly styled Nissan Murano .\nBut the Edge was in line for a styling facelift after a complete makeover a couple years ago .\nSo Ford decided to ditch the SUV 's relatively tame Sport edition and give it the full ST treatment .\n`` We sold over 140,000 Edges last year , a fair portion of those were Edge Sports , but this really takes that to another level , '' Nair said .\nBesides its track-tested suspension and race-ready seats , the Edge ST 's looks take a dark turn , with a honeycomb black gloss grill and matching 21-inch shiny black rims with big red brake calipers peeking through .\nThe SUV 's designer is not worried about turning off traditional buyers with the racing touches .\n`` The female buyers are asking , ` Make it more aggressive , so I will feel more confident in this vehicle , ''' Gilmore said .\nTop-Heavy Tweaks\nThe engineers also had to overcome a challenge not associated with low-slug sedans -- top-heavy SUVs have a tendency to tip over if they fly through a corner too fast .\nFord tuned the Edge ST to compensate for its high ride by `` managing the all-wheel drive system and electronically creating a little bit of sportiness without getting ourselves into any issues on the center of gravity , '' Nair said .\n`` Consumers may not know how to drive a performance SUV versus a car , '' said Jeff Schuster , an analyst with researcher LMC Automotive .\n`` There probably needs to be some education so we do n't see people sitting on the side of their vehicle . ''\nThe Edge ST goes on sale this summer , but Ford has n't revealed its price yet .\nThe Edge Sport it replaces starts at $ 40,675 and Cox 's Brauer expects Ford to cash in on the upgrades .\n`` SUV buyers traditionally have been willing to spend more money , '' Brauer said .\n`` Let 's be honest , the massive shift toward SUVs and trucks allowed all the car companies to make more money selling fewer cars last year . ''","token_count":939}
-{"event_count":47,"event_sentiment_count":1,"id":"f14","participant_count":98,"relation_count":17,"sentence_count":35,"sentiment_count":17,"split":"test","text":"Ford Is The Next GE And Shorts Should Be Salivating\nAfter writing extensively for Forbes on GE 's plummet in the past few months , it was striking that both GE and Ford delivered negative guidance to the Street last Tuesday .\nFord 's truth bomb -- delivered at the Deutsche Bank Global Auto Industry Conference -- was particularly destructive .\nFord CFO Bob Shanks noted that 2018 earnings were likely to be in the range of $ 1.45 - $ 1.70 versus a preliminary read on 2017 EPS of $ 1.78 ( full results are due Wednesday . )\nSo , in the midst of the Trump jump in the U.S. and data showing an increasing rate of growth in almost every other major global economy , Ford is predicting -- at the midpoint of its guidance range -- an EPS decline of 10 % in 2018 .\nHow does that happen ?\nHow can things be going so well in other sectors and not for Ford ?\nAccording to Ford management , the problem -- much like the ones at GE -- is due to factors outside management 's control .\nShanks used the phrase `` some of the unknowns or the unexpected '' to describe the negative impacts Ford is feeling from commodities and currencies .\nI have been listening to Ford conference calls for 25 years , beginning with my days as a cub analyst at Lehman in 1992 , and I am quite certain that Ford 's cars and trucks have always been made from steel , aluminum and other metals , and that Ford has always transacted business in many countries , thus incurring currency risk .\nSo , why are commodities and currencies causing such an earnings shortfall now ?\nTo be fair , Shanks started flagging increasing commodity prices as a risk to Ford 's earnings in his comments on F 's April 2017 earnings call , but what is Ford management doing to mitigate the impact of rising prices for metals , especially steel ?\nNot much .\nThis is Ford 's official transcription of Shanks ' response to an analyst question at the DB conference :\n`` We do n't hedge as many because our analysis had indicated that there are some currencies that provide a natural hedge to a metal like copper in Australian dollar I think , tend to move in opposite directions .\nSo we actually do n't hedge as much on currencies or commodities as a result .\nWhat we have seen in more recent [ ph ] short time ( 01:04:00 ) is those natural hedges are going in the opposite direction , which are not working .\nWe 're actually getting the bad news on the exchange , we 're getting the bad news on commodities . ''\nTranslation : our margins are suffering and we are not doing anything about it .\nBut is n't this just a pause as Ford transforms itself into a mobility company with leading positions in autonomous vehicles ( AVs ) and battery-electric vehicles ( BEVs ) ?\nBased on the Deutsche Bank presentation ( available here ) , Ford has not yet flushed out its strategy for AVs and BEVs .\nIn fact , Ford would seem to be years behind its competition , especially crosstown rival General Motors .\nRead GM 's masterful 91-page manifesto on its strategy for AVs and BEVs ( available here ) and then read Ford 's , and the message is clear .\nIn fact , it seems to me that the only hope remaining for Ford shareholders is yield support from the company 's dividend .\nIf you read the preceding paragraphs and still want to own Ford shares for the long term , I applaud your courage and will congratulate you on receiving the company 's recently declared special dividend of $ 0.13 in addition to your regular quarterly dividend of $ 0.15 % on March 1st .\nThat said , remember that GE also used to be a dividend stock .\nEven now , with a current yield of 3 % GE is still yielding more than full percentage point more than the S&P 500 .\nGE shares are also down 45 % in the past 12 months .\nAm I saying that could happen to Ford shares ?\nYes .\nIf Ford earns $ 1.45 in 2018 -- the low end of CFO Shanks ' guidance range -- and the shares fall below their recent P/E of range of 6.5-7.5 x , then that that would put a downside valuation on F shares of $ 7.00 - $ 7.50 , or about 40 % below where they trade today .\nIn prior Forbes columns , I predicted Ford would be `` the worst stock of the year '' in 2015 and 2016 , and F shares massively underperformed the S&P 500 in both periods .\nThis is is a slightly different call , though , I 'm looking for a sharp decline in Ford shares this year , not just relative underperformance , and that means F is an out-and-out short .\nYou may want to hedge a Ford short with a long GM position to have exposure to a componay that actually has a fully-defined strategy for autonomous vehicles and electrified powertrains , but I do n't think you will need to .\nI believe that Ford shares are a better candidate for shorting now than at any time since the fall of 2008 .","token_count":972}
-{"event_count":39,"event_sentiment_count":3,"id":"f15","participant_count":114,"relation_count":13,"sentence_count":54,"sentiment_count":39,"split":"test","text":"Ford Is At A Crossroad Of Danger And Opportunity In China\nIt is no secret that Ford is facing a minor crisis in China .\n-- January sales were down 18 percent year-on-year . ( Sales at rival GM were up 15 percent )\n-- Ford China CEO Jason Luo - in the position just 5 months - has resigned\n-- Momentum outside of Lincoln has basically stalled , while partner tensions simmer .\nAll is not lost , however .\nFord leaders might be relieved to learn that the Chinese word for crisis is composed of two characters , 危 and 机 。\nThe first , wei , means danger .\nThe second , ji , means opportunity .\nNegatives and positives reside side by side .\nFord 's crisis , indeed , contains elements of both danger and opportunity .\nFirst , the dangers : Ford 's China sales could continue to stall for months .\nMain culprit : Product offerings are getting old and the new metal will not arrive until the end of 2018 .\nFord 's slump , which started in 2017 , also feels heavier because it contrasts with the 5 previous years of record growth .\nBetween 2012 and 2016 , Ford doubled it sales to 1.2 million .\nWhat else is causing a feeling of foreboding ?\nVastly improved Chinese competitors like Geely and Great Wall are gobbling up fresh market share .\nIn 2017 , Geely even outsold Ford .\nA conventional response would be to hold one 's ground and wait for new vehicles .\nLike an army unit under siege calling for backup .\nThis approach might have worked in the past .\nBut times have changed .\nChinese competitors pose a real danger .\nTo regain strength , Ford must try bold new things .\nLucky for Ford , China provides fertile ground for experimentation - and opportunity .\nFord is already starting to shape a new and different kind of operation .\nTwo examples :\nElectrics .\nFord surprised the industry when it announced a new joint venture with Zotye , a private electric vehicle maker .\nFord and Zotye will start producing their first EVs at the end of 2018 or early 2019 .\nChina leads the world in electric vehicle sales ( 777,000 in 2017 ) .\nBeijing technocrats want that number to reach 1 million this year and then zoom to 6 million by 2025 .\nAn earlier , less nimble , version of Ford would have studied the various partner options for a couple of more years before settling on Zotye as a partner .\nPeter Fleet , president of Ford Asia Pacific , says we can take the electric vehicle projections to the bank .\n`` Their [ EV sales ] are going to be even bigger than you think , '' he says .\n`` The government makes it clear they want [ their EV industry ] to have great , global success . ''\nTech Titan Tie-Up .\nIn December , Chairman Bill Ford and CEO Jim Hackett traveled to the headquarters of Alibaba and signed an agreement with Jack Ma to explore new and inventive ways to deliver cars to customers .\nAlibaba , the Amazon/Ebay of China , is the PRC 's second most valuable company after Tencent .\nGetting together with Alibaba is important because it opens up direct channels to Chinese consumers and inventive ways to sell cars .\nHere are two other promising areas for Ford :\nSmart Cities : Chinese cities are home to some of the worst traffic jams and air pollution on the planet .\nDidi , the Uber of China , is working closely with major cities to improve overall traffic flows and transportation efficiencies .\nFord 's new global clarion call - smart cars for smart cities - looks like a timely solution for China 's problems .\nLincoln .\nChinese consumers bought an astounding 2.3 million luxury cars in 2017 .\nGerman automakers - Mercedes , BMW , Audi and Porsche - dominate with 80 % of the segment .\nBut Lincoln jumped its own sales up to 55,000 in 2017 , from 11,000 in 2015 .\nPlenty of upside there .\nConfronted with slumping sales , Ford is right to venture out and take risks .\nMore daring moves into the electric , ride-sharing and direct distribution arenas are the only way for the Blue Oval to prevail in China over the long term .\nFord needs to be a fundamentally different company in China -- there is no Ford F-150 profit machine to lean on here .\nSo , Ford in China should heed its own advice as it steps into the fresh territory of electrics , autonomy and smart city transportation : Go Further .\nWithin a crisis reside the seeds of opportunity .","token_count":837}
-{"event_count":33,"event_sentiment_count":2,"id":"jnj04","participant_count":57,"relation_count":15,"sentence_count":19,"sentiment_count":10,"split":"test","text":"Johnson & Johnson Earnings : When Strong Is n't Strong Enough\nWhen a stock is at all-time highs and sitting on major gains over the past year , it has to be perfect , not just on the surface but underneath as well .\nThough Johnson & Johnson ( JNJ ) reported better-than-expected fourth-quarter earnings Tuesday morning and a strong 2018 outlook , there was enough for the market to nitpick .\nInvestors sent the stock down 3.1 % , to $ 143.56 , marking its largest decline since last April and making it the second-worst performer in the Dow Jones Industrial Average .\nNot much has changed for the health-care conglomerate , analysts said , but there was enough for investors to take some profits .\n`` In this market , a company -- not just J&J -- needs to deliver clean , perfect quarterly results and guides , not just on the headline number but everything that goes into it , '' Glenn Novarro , an analyst at RBC Capital Markets , tells Barron 's .\nHe still has an outperform rating on the stock .\nHere are some of the issues : Operating margins for the quarter came in lighter than some expected , in part because of higher research-and-development spending that was offset by a lower tax rate .\nFor 2018 , the company said it expected about one percentage point of improvement in operating margins -- lighter than some investors were expecting , Novarro says .\nCurrency and tax reforms were a major driver for the strong results and full-year outlook .\nBut Leerink Partners analyst Danielle Antalffy says the company is not going to get credit for `` nonoperational '' guidance .\nPlus , organic sales growth projections for this year , based on management 's comments , suggested the momentum seen in the fourth quarter would n't continue at the same pace , even though J&J sees growth accelerating for the full-year , Antalffy says .\nSales of the company 's cancer drug Imbruvica were also lighter than expected ( $ 522 million instead of consensus expectations of $ 537 million ) .\nDarzalex , another J&J cancer drug , had strong global sales , but that number masked a lighter showing in the U.S. that was offset by strength abroad .\nBoth are drugs that analysts think represent the company 's future growth prospects .\nPotential note for AbbVie ( ABBV ) : J&J cited increased uptake globally and higher market share , wrote Credit Suisse analyst Vamil Divan .\nTony Butler , an analyst at Guggenheim Partners , also noticed several questions from analysts about potential generic competition for the company 's blockbuster prostate-cancer drug Zytiga after a court invalidated its patent last week .\n`` It could be a banana peel we slip on , but in the big scope of things , you have to shrug your shoulders , '' says Butler , who characterized the results as strong overall .\nLesson to companies : Strong outlooks built on lower taxes and favorable currency exchange rates may not be enough to cut it in this market .","token_count":551}
-{"event_count":40,"event_sentiment_count":1,"id":"jnj05","participant_count":75,"relation_count":6,"sentence_count":24,"sentiment_count":10,"split":"test","text":"Where 's The Tylenol ?\nJ&J Disappoints And Frustrates\nAt first glance , Johnson & Johnson seemed to deliver the goods .\nThe health-care conglomerate unveiled fourth quarter financial results and full-year 2018 guidance early Tuesday that beat expectations .\nDuring today 's conference call , it also announced plans to immediately bring back $ 12 billion in overseas cash , taking advantage of the lower tax rate on repatriated earnings and cash under the new tax law .\nBut as the day progressed and analysts dug below the headline numbers , those results lost their shine , and investors headed for the exit .\nDown 4.2 percent to end the day at $ 141.86 a share , J&J 's stock was one of the Dow Jones Industrial Average 's biggest laggards .\nIt 's a reminder that in this market , if a stock is bouncing off an all-time high , having climbed 30 percent over the previous 12 months , it ca n't afford to miss the mark .\nStrong quarterly results and guidance spurred by a positive currency impact and tax reform do n't cut it .\nAt $ 1.74 , per-share , fourth quarter operating profit beat the $ 1.72 expected by analysts .\nSales jumped 11.5 percent , spurred by pharmaceutical sales and last year 's $ 30 billion acquisition of Swiss biotech giant Actelion .\nBut profit margins came in lighter than expected .\nWithout the impact of a weaker U.S. dollar , global sales grew at a 9.4 percent clip .\nRemove acquisitions and divestitures from the mix , and organic sales rose 4.2 percent .\n`` ... the story in 4Q is the surprisingly weak gross margin missed by -300 bps with op ex management and a materially lower tax rate -1,000 bps vs. consensus helping to drive the modest EPS beat , '' wrote Barclays analyst Geoffrey Meacham in a note published today .\n`` ... We think that JNJ is poised to deliver a good 2018 , but we await a better entry point given the punchy valuation . ''\nHeadwinds continued to batter J&J 's biggest-selling drug in the quarter .\nSales of the blockbuster inflammation therapy Remicade sagged 10.5 percent in the face of rising competition .\nSales of Stelara , Invokana and other prominent drugs fell short of analyst expectations .\nAdding to worries : the threat of generic competition facing J&J 's blockbuster cancer therapy Zytiga after a court ruling last week invalidated its patent .\nAs for J&J 's full-year outlook , the company expects to earn between $ 8 a share and $ 8.20 a share this year , as sales climb 3.5 percent to 4.5 percent .\nBut organic sales growth comes in at 2.5 percent to 3.5 percent for the current year .\nJ&J expects operating margins to expand this year .\nIt has also promised to spend more on research and development .","token_count":525}
-{"event_count":25,"event_sentiment_count":0,"id":"jnj06","participant_count":51,"relation_count":8,"sentence_count":17,"sentiment_count":1,"split":"test","text":"What To Expect From Johnson & Johnson In 2018\nWe have created an interactive dashboard on Johnson & Johnson 's expected financial performance in 2018 .\nYou can adjust the revenue and margin drivers to see the impact on the company 's performance .\nOverall , we forecast revenue growth of around 5 % and EPS of just under $ 8 for 2018 .\nWe expect J&J 's Pharmaceuticals revenue to grow by 8 % in 2018 , primarily led by oncology drugs , along with cardiovascular , metabolism & other drugs .\nWithin oncology , Imbruvica and Darzalex have seen solid growth of late , and we expect this trend to continue in 2018 .\nIn the Consumer Healthcare business , we expect revenues to grow in low-single-digits , as skin care revenue growth will likely offset any pressure on baby care and other revenues .\nNote that baby care and other revenues have seen recent declines amid competition from private labels .\nMeanwhile , we forecast Medical Devices revenues to grow in low-single-digits .\nThe company saw a strong uptick in vision care sales in 2017 due to the AMO acquisition , and we expect the company to benefit from that in 2018 as well .\nJ&J 's Net Income Margin has been around 26 % on average over the past few years , and we expect that to remain relatively steady in 2018 .\nGiven our revenue growth forecast , we forecast around 5 % growth in net income in 2018 to about $ 21 billion .\nOur net income forecast of $ 21 billion translates into EPS of about $ 7.80 in 2018 .\nWe estimate a price-earnings multiple of around 19 for J&J , which is below many pharma industry multiples , reflecting the risk of biosimilars impacting the company 's future growth .\nThis translates into a price estimate of $ 149 for Johnson & Johnson 's stock , which is around 15 % above the current market price .\nWhat 's behind Trefis ?\nSee How it 's Powering New Collaboration and What-Ifs","token_count":364}
-{"event_count":27,"event_sentiment_count":0,"id":"nem04","participant_count":81,"relation_count":12,"sentence_count":22,"sentiment_count":3,"split":"test","text":"Analyst Insight : Is Newmont Mining Warming Up for a Good 2018 ?\nAnalyst ratings\nAlong with Agnico Eagle Mines ( AEM ) , Newmont Mining ( NEM ) has a high percentage of `` buy '' ratings among the senior miners at 61 % .\n39 % of the analysts that cover NEM have given it `` hold '' ratings .\nThe stock does n't have any `` sell '' ratings .\nIts target price implies an upside of 11.6 % from its current market price of $ 38.3 .\nThere has n't been much change in analysts ' ratings for NEM 's stock in the last few months .\nIn 2017 , apart from an upward revision in its unit costs for the next two years , the company has performed well overall .\nUpgrades\nNewmont Mining 's stock has not seen any rating change from analysts since July 2017 when Argus upgraded it from `` hold '' to `` buy '' with a target price of $ 43 .\nThe firm 's analyst had reasoned that the stock 's valuation at the time was attractive and expected the company to benefit from the positive fundamentals of the gold industry ( GLD ) .\nOn June 15 , 2017 , Raymond James initiated coverage of NEM with an `` outperform '' rating and a target price of $ 43.00 .\nRaymond James was also bullish on Kinross Gold ( KGC ) , Agnico Eagle Mines ( AEM ) , Alamos Gold ( AGI ) , and Detour Gold ( DRGDF ) .\nRecent rating changes\nBMO Capital Markets reiterated its `` buy '' rating on NEM on December 12 , 2017 .\nThe firm has a target price of $ 43 on the stock .\nSimilarly , Royal Bank of Canada ( or RBC ) Capital Markets also reaffirmed its `` hold '' rating on Newmont on December 5 , 2017 .\nOn October 31 , 2017 , FBR & Co. reiterated its `` hold '' rating on NEM 's stock with a target price of $ 40 .\nFBR analyst Lucas Pipes updated his estimates for the company after its 3Q17 results and updated 2017 guidance .\nTD Securities also reaffirmed its `` hold '' rating for Newmont 's stock on October 19 , 2017 , with a target price of $ 42 .\nOn the same day that TD Securities boosted its target price for NEM , Scotiabank also upped NEM 's target price from $ 43.50 to $ 55.00 .\nScotiabank had previously upgraded Newmont from `` sector perform '' to `` outperform '' on July 5 , 2017 .","token_count":477}
-{"event_count":24,"event_sentiment_count":0,"id":"nem05","participant_count":54,"relation_count":10,"sentence_count":13,"sentiment_count":13,"split":"test","text":"Newmont-Barrick Race for Top Gold Crown Comes Down to a Decimal\nGold geeks will have to wait for Newmont Mining Corp. to disclose one more decimal place to see if it overtook Barrick Gold Corp. as the world 's biggest bullion producer .\nOn Wednesday , the Greenwood Village , Colorado-based company said it produced 5.3 million ounces of gold last year , essentially tying the output of Toronto-based Barrick , which last week reported 5.32 million ounces .\nWhether that 's enough for it to squeak ahead of Barrick depends on the rounding .\nA spokesman for Newmont , Omar Jabara , said the more precise number will be given Thursday when the company posts fourth-quarter earnings and updates its production guidance .\nThe last year Newmont produced more gold than Barrick was 2005 .\nIn 2006 , Barrick vaulted ahead after buying Placer Dome Inc. and has maintained an output lead ever since , although Newmont has moved well ahead by market value .\nNewmont executives have increasingly been talking up the miner 's growth prospects , a contrast to Barrick , which has spoken more about margin improvements and debt-reduction efforts .\nNewmont has also stressed the need to reward shareholders with higher dividends , and on Tuesday doubled the quarterly payout to 14 cents .\nAll of this may have helped Newmont shares outperform , widening the value gap between the two companies .\nNewmont is pursuing greenfield exploration around the world and has added production with nine mine expansions in five years , as well as the purchase of the Cripple Creek & Victor mine in Colorado .\nAll told , the company has added more than 2 million ounces of gold production , Goldberg told investors in December .\nThe midpoint of Newmont 's current guidance suggests it will produce more gold than Barrick from 2018 to 2022 .","token_count":322}
-{"event_count":27,"event_sentiment_count":0,"id":"nem06","participant_count":65,"relation_count":10,"sentence_count":18,"sentiment_count":7,"split":"test","text":"Newmont Mining Is Investors ' Gold Stock to Buy\nGold mining stocks are not cheap , but if you want to own gold , Newmont Mining Corporation ( NEM ) is a good choice , as it best tracks the performance of Comex gold futures .\nNewmont is not cheap fundamentally , as its P/E ratio is 200.67 , and the dividend yield is anemic at 0.77 % .\nInvestors should have 10 % of their portfolio invested in gold-related choices , and Newmont is one such choice .\nThe stock closed Tuesday at $ 38.39 , up 2.3 % year to date after falling into correction territory at its 2018 low of $ 35.98 on Feb. 9 .\nNewmont shares are up 6.7 % since then .\nThe stock peaked at $ 42.04 on Jan. 24 and is currently 8.7 % below that high .\nAnalysts expect Newmont Mining to post earnings per share of 40 cents when it reports results before the opening bell on Thursday , Feb. 22 .\nSome analysts say that gold production should continue to improve in Africa and North America .\nMeanwhile , copper is another metal mined by Newmont that is projected to have year-over-year gains .\nThe daily chart for Newmont Mining shows that it has been above a '' golden cross '' since Aug. 10 , when the stock closed at $ 36.44 .\nA `` golden cross '' occurs when the 50-day simple moving average rises above the 200-day simple moving average , indicating that higher prices lie ahead .\nThe horizontal lines show that the stock held its semiannual value level of $ 36.15 , when it held its 200-day simple moving average of $ 36.20 .\nThe stock is above its 50-day simple moving average of $ 38.10 , with its monthly risky level at $ 40.76 .\nThe weekly chart for Newmont Mining will be negative if the stock closes Friday just below its five-week modified moving average of $ 38.43 .\nThe stock is above its 200-week simple moving average at $ 29.06 , which is also the `` reversion to the mean , '' last tested during the week of April 15 , 2016 , when the average was $ 28.91 .\nThe 12 x 3 x 3 weekly slow stochastic reading is projected to end this week at 60.45 , down from 65.88 on Feb. 16 .\nGiven these charts and analysis , investors should buy Newmont Mining shares on weakness to my semiannual value levels of $ 36.15 and reduce holdings on strength to my monthly risky level of $ 40.76 .","token_count":463}
-{"event_count":31,"event_sentiment_count":1,"id":"wmt05","participant_count":53,"relation_count":16,"sentence_count":13,"sentiment_count":2,"split":"test","text":"Walmart Stock Nears Key Support After Earnings Miss\nWalmart Inc. ( WMT ) shares have fallen more than 12 % since the beginning of the week after the company reported worse-than-expected fourth quarter financial results .\nRevenue rose 4.1 % to $ 136.3 billion -- beating consensus estimates by $ 1.39 billion -- but earnings per share hit only $ 1.33 and missed consensus estimates by four cents per share .\nThe company 's full-year profit guidance also came in at $ 4.75 to $ 5.00 per share , below expectations of $ 5.13 per share .\nAside from the lackluster guidance , the company 's e-commerce growth came in at just 23 % , which was sharply lower than the growth of roughly 40 % seen in past quarters .\nManagement primarily attributed the slower growth to the Jet.com acquisition that added scale but anticipates the growth rate to ramp back up to the 40 % range after the first quarter .\nFull-year e-commerce sales remain up 44 % versus the prior year .\nFrom a technical standpoint , the stock broke down from trendline support earlier this month , rebounded to the pivot point and fell again to key support levels .\nThe relative strength index ( RSI ) appears oversold at 31.71 , but the moving average convergence divergence ( MACD ) remains in a bearish downtrend .\nThese two technical indicators suggest that the stock could see some consolidation and a possible move even lower if the trend reverses in the longer term .\nTraders should watch for some consolidation above trendline support levels after closing the gap dating back to mid-November .\nIf the stock breaks down from these levels , it could reach the 200-day moving average at around $ 86.21 or reaction lows at around $ 77.50 .\nIf the stock rebounds , traders should watch for a move to S1 support and the 50-day moving average at around $ 100.00 on the upside .","token_count":341}
-{"event_count":30,"event_sentiment_count":1,"id":"wmt06","participant_count":56,"relation_count":9,"sentence_count":21,"sentiment_count":12,"split":"test","text":"Goldman expects Wal-Mart 's fortunes to improve alongside the consumer ’s\nGoldman Sachs upgraded Wal-Mart to buy and put the retail giant on its Conviction List\nGoldman Sachs thinks 2018 will be a good year for consumers , which means it will also be a good year for Wal-Mart Stores Inc. .\nGoldman upgraded Wal-Mart to buy from neutral and put the retail giant on its Conviction List .\nIts price target was moved up $ 2 to $ 117 .\nWal-Mart shares are up 1.1 % in Thursday trading .\n`` We expect growth in disposable income to reaccelerate , helping to sustain a new plateau for consumption , with more consistent growth across verticals than we have noted in the recent low , stable growth phase of the recovery , '' wrote Goldman Sachs analysts .\n`` This forecast reflects accelerating wage and income dynamics , and additional help from tax reform . ''\nGoldman expects tax reform to give the biggest boost to consumption capability to those whose returns reflect an adjusted gross income of $ 200,000 to $ 1 million .\nBut those in the $ 40,000-to - $ 200,000 range should also feel a `` substantial '' impact .\nAmong the areas where Goldman thinks there will be year-over-year same-store sales increases are apparel , food and casual dining , `` and outright like-for-like declines at any point in the year only for department stores . ''\nDespite the largely bullish outlook , Goldman analysts leave room for risk , noting that retail is still vulnerable to disruption .\nOn Wednesday , Instinet analysts noted the impact that a delayed start to the tax return season could have on certain retailers .\nWeather is also frequently cited by retail executives as a factor that can both help or hurt results .\nThe consumer sector has become more bullish since late summer , and upbeat holiday season results suggest that some of the hurdles that have faced brick-and-mortar and traditional retailers may be lessening .\nWal-Mart has the advantage of being `` very much in control of its own destiny , '' Goldman writes .\n`` The firm 's strategic positioning remains compelling -- selling consumables to middle-income consumers in small markets , '' the note said .\nAmong the key downside risks analysts highlight are competition from Amazon.com Inc. , discounters and traditional grocers .\nLast week , Wal-Mart said it would use savings from the new tax reform law to offer a series of new associate benefits , including a raise in the minimum wage to $ 11 and bonuses of up to $ 1,000 , and expanded maternity and parental leave allowances .\nWal-Mart is expected to announce fourth-quarter earnings on Feb. 20 .\nThe company 's stock is up 21 % for the last three months while the Dow Jones Industrial Average is up 12.8 % for the period .","token_count":510}
-{"event_count":26,"event_sentiment_count":0,"id":"wmt07","participant_count":77,"relation_count":13,"sentence_count":19,"sentiment_count":9,"split":"test","text":"Walmart 's Meal Kits Are Not The Solution To Fight Amazon\nThe nation 's largest food retailer announced this morning that it would roll out its own version of meal-kits to 2,000 of its stores this year .\nCurrently Walmart offers the kits , which retails from $ 8 to $ 15 and serves two people , in more than 250 stores .\nSpeculation in early February by Gabelli & Co. analyst Matthew Trusz was that Walmart was looking to acquire Blue Apron actually raised the troubled company 's stock price by almost 7 percent on February 2nd ; but it appears that Walmart has decided to enter the meal kit business on its own ( at least for now ) .\nAccording to Nielsen 's latest report just released today , The Meal Kit Opportunity , while the majority of meal kits are purchased from online subscription services , in-store meal kit sales grew rapidly in the past year , up 26.5 % to $ 154.6 million in sales , posting growth of more than 26 % year-over-year .\nOverall , the report states , 9 % of Americans say they 've purchased a meal kit in the last six months and 25 % of consumers say they would consider trying a meal kit in the next six months .\nIn December of 2017 , USDA published a report on Millenials eating and food purchasing habits and found that this generation , which now makes up roughly 26 percent of the total US population , eat out more frequently than other generations , spend more on prepared foods that require minimal preparation , spend less time preparing food at home and spend less money on foods to consume at home .\nThe report also found that nearly two-thirds of Millenials purchased prepared foods from limited service restaurants every week .\nTyler Lehr Walmart 's senior vice president of Deli Services said in the statement announcing the roll out of the program that `` more than 80 percent of Americans do n't know what they will have for dinner tonight .\nScrambling to find a meal `` puts pressure on a family . ''\nWalmart 's roll out also includes fresh One Step Meals that only require heating -- which may have even a larger potential than their Meal Kits .\nWhile Blue Apron , Hello Fresh , Martha & Marley and dozens of other DYI food kits tout that consumers enjoy the assembly process and how their kits make it so easy for the home cook to excel ; many people say they have already become bored with having to unpack all the ingredients and do all the preparation ; many also complain about the wasteful packaging that is required both in shipping and individually packaging each ingredient .\nSupermarkets across the nation have rushed to offer meal kits in their sotes that eliminate the extra packaging required for shipping and the need to pre-order multiple meals which in some cases are delivered in three days .\nIn 2005 `` meal assembly stores '' were all the rage and according to the now defunct Easy Meal Preparation Association there were over 1,400 of these stores throughout the US .\nToday Dream Dinners is one of the few chains to survive with 75 locations in 22 states .\nWhich makes one wonder if the Meal Kit trend might find itself in the same position .\nWalmart 's strategy comes on the heels of Amazon announcement that it will be delivering its own prepared foods and meal kits from its Whole Foods stores .\nWhile both companies have publicly said they are fierce competitors , this move by Walmart brings meal kits and prepared foods to a less affluent and larger base shopper than those who shop at Whole Foods .\nIt would not be surprising to also see these type of offerings start appearing in hipper packaging as they are a perfect fit for Walmart 's Jet.com business , which already sells a wide variety of Omaha Steak meat , seafood and side dish products .","token_count":702}

data/documents/test.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:dafb487f0fd5221389846ba091b4148a0f331c85f18b6de31956cda322c65ded
-size 63825

data/documents/train.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:b516f6c358792ff21e02da6cd12796ee3e74f9b3f8f1af531fde8f72cf23ae6b
-size 420702

data/documents/validation.parquet

@@ -1,3 +0,0 @@
-version https://git-lfs.github.com/spec/v1
-oid sha256:475104fd2234ddeb34a013a14dfcc831627d27563c8de954ac40c790518509fc
-size 64812