| { | |
| "paper_id": "L16-1025", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T12:03:36.077022Z" | |
| }, | |
| "title": "Coreference Annotation Scheme and Relation Types for Hindi", | |
| "authors": [ | |
| { | |
| "first": "Vandan", | |
| "middle": [], | |
| "last": "Mujadia", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "IIIT-H", | |
| "location": { | |
| "country": "India" | |
| } | |
| }, | |
| "email": "[email protected]" | |
| }, | |
| { | |
| "first": "Palash", | |
| "middle": [], | |
| "last": "Gupta", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "IIIT-H", | |
| "location": { | |
| "country": "India" | |
| } | |
| }, | |
| "email": "[email protected]" | |
| }, | |
| { | |
| "first": "Dipti", | |
| "middle": [], | |
| "last": "Misra", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "IIIT-H", | |
| "location": { | |
| "country": "India" | |
| } | |
| }, | |
| "email": "[email protected]" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "This paper describes a coreference annotation scheme, coreference annotation specific issues and their solutions through our proposed annotation scheme for Hindi. We introduce different co-reference relation types between continuous mentions of the same coreference chain such as 'Part-of', 'Function-value pair' etc. We used Jaccard similarity based Krippendorff's' alpha to demonstrate consistency in annotation scheme, annotation and corpora. To ease the coreference annotation process, we built a semi-automatic Coreference Annotation Tool (CAT). We also provide statistics of coreference annotation on Hindi Dependency Treebank (HDTB).", | |
| "pdf_parse": { | |
| "paper_id": "L16-1025", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "This paper describes a coreference annotation scheme, coreference annotation specific issues and their solutions through our proposed annotation scheme for Hindi. We introduce different co-reference relation types between continuous mentions of the same coreference chain such as 'Part-of', 'Function-value pair' etc. We used Jaccard similarity based Krippendorff's' alpha to demonstrate consistency in annotation scheme, annotation and corpora. To ease the coreference annotation process, we built a semi-automatic Coreference Annotation Tool (CAT). We also provide statistics of coreference annotation on Hindi Dependency Treebank (HDTB).", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "There has been considerable research for coreference annotation on various languages (English (Hovy et al., 2006) , French (Mitkov et al., 2000) , Spanish (Recasens et al., 2007) , Czech (Nedoluzhko et al., 2013) , etc), on diverse domains like newspaper texts, bio-medical journals, etc.", | |
| "cite_spans": [ | |
| { | |
| "start": 94, | |
| "end": 113, | |
| "text": "(Hovy et al., 2006)", | |
| "ref_id": "BIBREF7" | |
| }, | |
| { | |
| "start": 123, | |
| "end": 144, | |
| "text": "(Mitkov et al., 2000)", | |
| "ref_id": "BIBREF9" | |
| }, | |
| { | |
| "start": 155, | |
| "end": 178, | |
| "text": "(Recasens et al., 2007)", | |
| "ref_id": "BIBREF14" | |
| }, | |
| { | |
| "start": 187, | |
| "end": 212, | |
| "text": "(Nedoluzhko et al., 2013)", | |
| "ref_id": "BIBREF11" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "Coreference annotation is a time consuming, challenging and an expensive task. To overcome these challenges, various coreference annotation tools have been developed like CorefDraw (Harabagiu et al., 2001 ), GATE (Cunningham et al., 2002) , PALinkA (Orasan and Sb, 2000) , MMAX2 (M\u00fcller and Strube, 2001 ) and BART (Versley et al., 2008) . All these provide text based visualization for annotation. Also there has been considerable work on coreference type relations, like, (Recasens et al., 2010) presented a typology of Near-Identity Relations and motivated the need for a middle ground category between identity and non-identity in the coreference task.", | |
| "cite_spans": [ | |
| { | |
| "start": 181, | |
| "end": 204, | |
| "text": "(Harabagiu et al., 2001", | |
| "ref_id": "BIBREF6" | |
| }, | |
| { | |
| "start": 213, | |
| "end": 238, | |
| "text": "(Cunningham et al., 2002)", | |
| "ref_id": "BIBREF4" | |
| }, | |
| { | |
| "start": 249, | |
| "end": 270, | |
| "text": "(Orasan and Sb, 2000)", | |
| "ref_id": "BIBREF12" | |
| }, | |
| { | |
| "start": 279, | |
| "end": 303, | |
| "text": "(M\u00fcller and Strube, 2001", | |
| "ref_id": "BIBREF10" | |
| }, | |
| { | |
| "start": 315, | |
| "end": 337, | |
| "text": "(Versley et al., 2008)", | |
| "ref_id": "BIBREF16" | |
| }, | |
| { | |
| "start": 474, | |
| "end": 497, | |
| "text": "(Recasens et al., 2010)", | |
| "ref_id": "BIBREF15" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "For Indian languages, co-referentially annotated corpora are few in numbers and mostly for Hindi. As stated in (Dakwale et al., 2012) , most schemes which are meant to be for English are not applicable for Indian languages due to their free word order. (Dakwale et al., 2012) presents an anaphora annotation scheme and applied it on Hindi Dependency Treebank for limited set of pronominal categories, particularly for concrete anaphors. (Dakwale et al., 2012) used key-value pair attributes on anaphor chunk-head to represent its referent(s). Compared to English coreference annotation (and its representation), very little work has been done on Indian languages. This paper tries to fill that gap by describing our coreference annotation scheme for Indian Languages with Hindi examples. We also discuss issues related to coreference annotation specific to Hindi such as distribution of markable span, level of annotation, concept of chain, type relation between mentions(referential entities), full and partial membership of mentions in a chain, multi-chain membership of mentions in chains and their representation in SSF (Bharati et al., 2007) .", | |
| "cite_spans": [ | |
| { | |
| "start": 111, | |
| "end": 133, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 253, | |
| "end": 275, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 437, | |
| "end": 459, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 1124, | |
| "end": 1146, | |
| "text": "(Bharati et al., 2007)", | |
| "ref_id": "BIBREF3" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "We point out several issues in the earlier annotation schemes and describe a consistent solution to those issues.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "Co-referentially annotated corpora were, traditionally, not annotated to capture the degree of relations between the referential entities. Our scheme also includes the degree of relation between continuous referential entities of the same chain into different relation types.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "This includes relations such as 'Part-of', 'Instance-of', 'Function-value pair' etc. These relation types would help various applications like question-answering, summarization, etc.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "The structure of the paper is as follows. Section 2 describes possible co-referential expressions in Hindi. Section 3 describes annotation scheme and design issues and their solutions. While in sections 4, 5, 6 we discuss relation types between mentions, inter annotator agreement study and coreference annotation tool(CAT) respectively. Section 7 describe the statistics of coreference annotated corpora (Hindi dependency treebank). We conclude the paper in section 8.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "When expressions refer to the same entity, they are said to be in co-referring relations. These expressions (mention or referential) can be anaphors, nominal sequences or verb-nominal sequences. For Hindi, we categorize anaphors according to pronominal forms and theirs reference types. As Indian languages exhibit relatively free word order, referentials (mentions) can also be, at times, fragmented. Also mentions can match a chunk, smaller (only a part) than a chunk or can be larger than a chunk. For our purpose noun sequences were divided into two types i.e. definite noun sequences and indefinite noun sequences. We took named entities, abbreviations, titles of named entities, numerals, etc in definite noun sequences. In indefinite noun sequences we included potential referential entities other than which are in definite noun sequences. For clausal or verb-clause references, we decided to annotate whole clause or sentence with all its attributes ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference in Hindi", | |
| "sec_num": "2." | |
| }, | |
| { | |
| "text": "\u092e\u094b\u0939\u093e\u0932\u0940 Mohali - - m1%1 \u092e\u094b\u0939\u093e\u0932\u0940 Mohali :m1 - - - \u0915\u093e 's \u2032 - - - - - - - \u092e\u094b\u0939\u0928 Mohan i1%1:t1 \u092e\u094b\u0939\u0928 Mohan :i1 - - m1 i1:t1 \u090f\u0915 one i2%0:t1 - - - - - - \u0905\u091a\u094d\u091b\u093e nice i2%0:t1 - - - - - - \u0932\u095c\u0915\u093e boy i2%1:t1 \u0932\u095c\u0915\u093e boy :i2 - - noun-noun:i1 \u0939\u0948 \u0964 is. \u0935\u0939 He i3%1:t1 \u0935\u0939 He :i3 - - - anaphora-C:i2 \u092e\u0941 \u0902 \u092c\u0908 \u092e\u0947\u0902 in_Mumbai \u0930\u0939\u0924\u093e \u0939\u0948 \u0964 live+PRES \u092e\u094b\u0939\u0928 Mohan i4%1:t1 \u092e\u094b\u0939\u0928 Mohan :i4 - - - noun-noun:i1", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference in Hindi", | |
| "sec_num": "2." | |
| }, | |
| { | |
| "text": "\u0906\u091c today \u0939\u0948 \u0926\u0930\u093e\u092c\u093e\u0926 Hyderabad \u0906\u092f\u093e \u0939\u0948 \u0964 came_. (sentential discourse) rather than only verb as referential element.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference in Hindi", | |
| "sec_num": "2." | |
| }, | |
| { | |
| "text": "This section describes our annotation scheme and compares it with (Dakwale et al., 2012) and MUC 1 schemes. Our coreference annotation scheme includes alltogether 7 fields. They are cref : This field represents the unique index for a mention, the unique index for a chain to which a mention belongs and the textual span of a mention [template-MentionId%(0/1):chainId], crefHead : This field represents the linguistic head of a mention, acrefmod : This field specifies the unique index for a modifier and its textual span, crefmod : This field is used to link a mention and its modifier with the unique modifier index, crefmodHead : This field represents the linguistic head of a modifier, crefType : This field specifies the type relation between mentions of the same chain, and crefChain-Head : This field is used to mark the head mention of the chain. fields using example (1). We can see in table (1) that mention indices i1, i2, i3, i4 in 'cref' tag are assigned to mentions \u092e\u094b\u0939\u0928 Mohan , \u090f\u0915 \u0905\u091a\u094d\u091b\u093e \u0932\u095c\u0915\u093e one\u2212nice\u2212boy , \u0935\u0939 he and \u092e\u094b\u0939\u0928 Mohan respectively. Unique chain index 't1' is assigned to each mention(reference of the same concept). Mentions' textual span is specified within 'cref' tag by either %0 and %1(indicate end of mention) value. 'crefHead' indicates the linguistic head for a mention. Here at \u0932\u095c\u0915\u093e boy , crefType='noun-noun' indicates that the mention \u090f\u0915 \u0905\u091a\u094d\u091b\u093e \u0932\u095c\u0915\u093e one\u2212nice\u2212boy has a 'noun-noun' relation with a mention \u092e\u094b\u0939\u0928 mohan (coreference type relations have been discussed in next section). 'acrefmod' and 'crefmod' are used to indicate the link between the mention and its modifier, the unique index in this case is 'm1'. crefChain-Head='i1:t1' is assigned on \u092e\u094b\u0939\u0928 mohan , which indicates that i1 is the head mention of the 't1' chain. Discontinuous parts of the same referential entity (single unit) captured by marking %0 and %1 with the unique mention index in 'cref' field. We point out the gist of coreference annotation schemes ((Dakwale et al., 2012) , MUC 2 [1]) and theirs weaknesses, observed by us. We overcome those limitations in our scheme.", | |
| "cite_spans": [ | |
| { | |
| "start": 66, | |
| "end": 88, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 1959, | |
| "end": 1982, | |
| "text": "((Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Annotation Scheme", | |
| "sec_num": "3." | |
| }, | |
| { | |
| "text": "while (Dakwale et al., 2012) 's scheme is only able to represent anaphors and their referents.", | |
| "cite_spans": [ | |
| { | |
| "start": 6, | |
| "end": 28, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Our annotation scheme accommodates coreference", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "2. Our scheme operates on lexicals to include referential mentions which are may be smaller or larger than chunk/phrase while (Dakwale et al., 2012) 's' scheme operates on chunks.", | |
| "cite_spans": [ | |
| { | |
| "start": 126, | |
| "end": 148, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Our annotation scheme accommodates coreference", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "3. In (Dakwale et al., 2012) scheme there is no provision to mark a relation between mentions of a chain, while MUC scheme 3 [1] has provision to mark limited relation types between mentions. In our scheme we used 'crefType' field for relation marking.", | |
| "cite_spans": [ | |
| { | |
| "start": 6, | |
| "end": 28, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Our annotation scheme accommodates coreference", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "4. Our scheme uses indeies to represent chains while in (Dakwale et al., 2012) (Dakwale et al., 2012 ) also can be solve by the flag value (0/1) used after % symbol in 'cref' tag, which indicates end of the lexical textual span of a mention. We can see from example (1) ( table 3) that we are capturing chain notion (points 4.-5.) by unique index (t) in cref field. We used 'crefType' field to capture the type relations between mentions (points 3.-6.).", | |
| "cite_spans": [ | |
| { | |
| "start": 56, | |
| "end": 78, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 79, | |
| "end": 100, | |
| "text": "(Dakwale et al., 2012", | |
| "ref_id": "BIBREF5" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Our annotation scheme accommodates coreference", | |
| "sec_num": "1." | |
| }, | |
| { | |
| "text": "To capture the degree of relations between referential entities, we mark relations between the continuous mentions of the same chain. We divided relations into following broader classes. They are anaphoric relations, strong identity relations, near identity relations and weak identity relations. All-together we categorize these relations into 21 sub-categories. Anaphoric relations include concrete, abstract, temporal relations while strong identity relations include the exact lexical match between mentions. Near identity relations include syntactic behavior between mentions like, noun-complement, apposition, abbreviation, etc. In weak identity relations we include relations like part-of where one mention partially refers to other mention. ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference relations", | |
| "sec_num": "4." | |
| }, | |
| { | |
| "text": "4.1.1. Anaphora-C : When pronominal mention has an individual or a concrete entity referent, for that we decided to mark \"Anaphora-C\" relation (C stands for Concrete) between those two mentions. If pronominal mention has another pronominal mention as a referent, then also we mark 'Anaphora-C' as a relation between those two pronominal mentions. In above example (2), a pronominal mention \u0905\u092a\u0928\u0947 his is referring to concrete entity \u092d\u093e\u091c\u092a\u093e \u0928\u0947 \u0924\u093e BJP_leader . Therefore, pronoun \u0905\u092a\u0928\u0947 his has a 'Anaphora-C' relation with \u092d\u093e\u091c\u092a\u093e \u0928\u0947 \u0924\u093e BJP_leader .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Anaphora relations :", | |
| "sec_num": "4.1." | |
| }, | |
| { | |
| "text": "This relation is similar to 'Anaphora-C' relation with only directional difference between a pronoun and its referent. i.e., first expression (here pronominal mention), that later co-refers with a more specific, second expression (referent mention) in the discourse. It is also known as a cataphora relation where the pronoun precedes its reference. Here 'RC' in 'Anaphora-RC' stands for Reverse Concrete.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Anaphora-RC :", | |
| "sec_num": "4.1.2." | |
| }, | |
| { | |
| "text": "When pronominal mention has an abstract/event entity as a referent, then we decided to mark it as \"Anaphora-E\" relation (E stands for Event) between two mentions. If a pronominal mention has another pronoun as referent then, also we mark 'Anaphora-E' as relation between those two pronominal mentions.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Anaphora-E :", | |
| "sec_num": "4.1.3." | |
| }, | |
| { | |
| "text": "( Other-then this, city has good junctions for bus, taxi and train.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Anaphora-E :", | |
| "sec_num": "4.1.3." | |
| }, | |
| { | |
| "text": "In above example (3), a pronoun \u0907\u0938\u0915\u0947 otherthen_this which is referring to abstract entity in previous sentence. Therefore, pronominal mention \u0907\u0938\u0915\u0947 otherthen_this has an 'Anaphora-E' relation with last sentential event.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Anaphora-E :", | |
| "sec_num": "4.1.3." | |
| }, | |
| { | |
| "text": "When a pronoun refers to time or time referring/representing an event or a clause in given discourse, for that we decided to mark \"Anaphora-T\" relation (T stands for Temporal) between those two mentions. Also, if pronominal mention has an another pronoun as a referent, then also, we decided to mark 'Anaphora-T' as a relation between those two pronouns. 4 (When i was asked that whose name you would like to take) i , then i I took my name", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Anaphora-T :", | |
| "sec_num": "4.1.5." | |
| }, | |
| { | |
| "text": "In above example (4), pronoun \u0924\u092c then is referring to time referring verb of previous sentence therefore a pronoun \u0924\u092c then has an 'Anaphora-T' relation with its previous sentence.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Anaphora-T :", | |
| "sec_num": "4.1.5." | |
| }, | |
| { | |
| "text": "Apart from above discussed pronominal reference types, in text there can be pronouns for which either no reference is specified or they do not have any reference. Like indefinite pronoun refers to something that is not definite or specific or exact. Indefinite pronouns include quantifiers (some, any, enough, several, many, much); universals (all, both, every, each); and partitives (any, anyone, anybody, either, neither, no, nobody, some, someone). Many of the indefinite pronouns can function as determiners. In Hindi \u0915\u094b\u0908 someone , \u0907\u0924\u0928\u093e this_much and \u0915\u0941 \u091b some are indefinite pronouns. In Hindi \u0915\u0941 \u091b some is used to indicate a portion or quantity of some entity. It is also use to indicate unspecified quantity of countable entities and unspecified portion of uncountable entities. The indefinite pronoun \u0915\u094b\u0908 someone used to indicate the absence of a portion or quantity of some entity. This indefinite pronouns are marked with 'Anaphora-others' relation.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Anaphora-others :", | |
| "sec_num": "4.1.6." | |
| }, | |
| { | |
| "text": "In this strong identity relation, we decided to mark only exact match (same lexicals/strings) and partial match (matched with entity head) mentions under 'strong identity relation' as they refer to same real world entity with same lexical patterns.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Strong Identity relation :", | |
| "sec_num": "4.2." | |
| }, | |
| { | |
| "text": "This only relation comes under strong Identity relation where two mentions have same lexical pattern and same mention head which are identical to each other.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Identity :", | |
| "sec_num": "4.2.1." | |
| }, | |
| { | |
| "text": "For example, \u0938\u093f\u091a\u0928 \u0930\u092e\u0947 \u0936 \u0924\u0947\u0902 \u0926\u0941 \u0932\u0915\u0930 Sachin_Ramesh_Tendulker and \u0938\u093f\u091a\u0928 Sachin ,\u0938\u093f\u091a\u0928 \u0930\u092e\u0947 \u0936 \u0924\u0947\u0902 \u0926\u0941 \u0932\u0915\u0930 Sachin_Ramesh_Tendulker and \u0924\u0947\u0902 \u0926\u0941 \u0932\u0915\u0930 Tendulker ,\u0930\u093e\u091c\u0917 \u0938\u0930\u0915\u093e\u0930 Ra jag_goverment and \u0938\u0930\u0915\u093e\u0930 Goverment are pairs of mentions which are identical to each other on referential as well as lexical bases but \u0930\u093e\u091c\u0917 \u0938\u0930\u0915\u093e\u0930 Ra jag_goverment and \u092f\u0942 \u092a\u0940\u090f \u0938\u0930\u0915\u093e\u0930 UPA_goverment , \u0930\u093e\u091c\u0917 \u0938\u0930\u0915\u093e\u0930 Ra jag_goverment and \u0930\u093e\u091c\u0917 Ra jag are certainly not.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Identity :", | |
| "sec_num": "4.2.1." | |
| }, | |
| { | |
| "text": "In these type of relations, two referents are representing same discourse entity although they may have different lexical pattern and also they can be in syntactic constructions. We divided this type into mainly 6 sub-types, base on their significance. Those types are discussed in following subsections.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Near Identity relations :", | |
| "sec_num": "4.3." | |
| }, | |
| { | |
| "text": "Apposition is a grammatical construction in which two mentions (a noun or noun phrase) are occurred subsequently, where one is serving to identify the other. Coreference-Apposition relation occurs when there is a proper noun and followed by its description, which also has an independent capability to replace previous proper noun for further reference in discourse. The mention pair having this property are said to be in apposition relation and we decided to mark this relation type between them. i.e. ,\u0938\u093f\u091a\u0928 \u0924\u0947\u0902 \u0926\u0941 \u0932\u0915\u0930, \u090f\u0915 \u092e\u0939\u093e\u0928 \u0916-\u0932\u093e\u095c\u0940 \u0939\u0948 . (Sachin Tendulker is one of the greatest player.)", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Apposition :", | |
| "sec_num": "4.3.1." | |
| }, | |
| { | |
| "text": "In this example \u0938\u093f\u091a\u0928 \u0924\u0947\u0902 \u0926\u0941 \u0932\u0915\u0930 Sachin_Tendulker is also explained as \u090f\u0915 \u092e\u0939\u093e\u0928 \u0916\u0932\u093e\u095c\u0940 one_o f _the_gretast_player , hence \u090f\u0915 \u092e\u0939\u093e\u0928 \u0916\u0932\u093e\u095c\u0940 one_o f _the_gretast_player is in apposition relation with \u0938\u093f\u091a\u0928 \u0924\u0947\u0902 \u0926\u0941 \u0932\u0915\u0930 Sachin_Tendulker . ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Apposition :", | |
| "sec_num": "4.3.1." | |
| }, | |
| { | |
| "text": "Jayandra_Saraswati is a proper noun and is followed by \u0915\u093e\u0902 \u091a\u0940 \u0915\u093e\u092e\u0915\u094b\u093f\u091f \u092a\u0940\u0920 \u0915\u0947 \u0936\u0902 \u0915\u0930\u093e\u091a\u093e\u092f\u0930\u094d (Shankaracharya of Kanchi Kamakoti union).", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "In above example (5), mention \u091c\u092f\u0947\u0902 \u0926\u0930\u094d \u0938\u0930\u0938\u094d\u0935\u0924\u0940", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "So according to our definition, \u091c\u092f\u0947\u0902 \u0926\u0930\u094d \u0938\u0930\u0938\u094d\u0935\u0924\u0940 Jayandra_Saraswati has 'Coreference-Apposition' relation with \u0915\u093e\u0902 \u091a\u0940 \u0915\u093e\u092e\u0915\u094b\u093f\u091f \u092a\u0940\u0920 \u0915\u0947 \u0936\u0902 \u0915\u0930\u093e\u091a\u093e\u092f\u0930\u094d (Shankaracharya of Kanchi Kamakoti union).", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "In above example (5), mention \u091c\u092f\u0947\u0902 \u0926\u0930\u094d \u0938\u0930\u0938\u094d\u0935\u0924\u0940", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Like Apposition, Noun Complement is a grammatical construction in which two nouns or/and noun phrases are placed side by side, and first noun phrase uses for showing designation or position of second noun phrase. In a way this relation has exactly reverse direction then apposition relation. A designation followed by a proper noun phrase make the Coreference-NounComplement relation. This relation is most commonly observed relation in Hindi dependency treebank . 6 In above example (6), two mentions, \u092a\u0942 \u0935\u0930\u094d \u0938\u0932\u093e\u092e\u0940 \u092c \u0947 \u092c\u093e\u091c f ormer_opener and \u0930\u092e\u0940\u091c Rameez are placed side by side and first noun sequence is showing designation for the second one. Therefore, mention \u0930\u092e\u0940\u091c Rameez has a 'Coreference-NounComplement' relation with \u092a\u0942 \u0935\u0930\u094d \u0938-\u0932\u093e\u092e\u0940 \u092c \u0947 \u092c\u093e\u091c f ormer_opener .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-NounComplement :", | |
| "sec_num": "4.3.2." | |
| }, | |
| { | |
| "text": "Like previous relation types, this relation type does not have linguistic significant but sentential construction for this type is very frequent in text. This relation type is used to show the relation between shortened form of a noun sequence and a noun sequence. Usually, shortened form consists of a letter or group of letters taken from the initial of individual words of a noun sequence. Specially for Hindi, where there can be two types of abbreviation, 1) noun sequence is in romanize Devanagari and it is abbreviated from their initials. 2) noun sequence is in Devanagari and its initials are taken in abbreviated form.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Abbreviation :", | |
| "sec_num": "4.3.3." | |
| }, | |
| { | |
| "text": "Reliance", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "(7) \u0930\u0932\u093e\u092f\u0902 \u0938 \u0928\u0947", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "\u0905\u092a\u0928\u0940 \u0917\u0948 \u0938 own_gas [ \u0928\u0947 \u0936\u0928\u0932 \u0925\u092e\u0930\u094d \u0932 \u092a\u0949\u0935\u0930 \u092a\u094d\u0932\u093e\u0902 \u091f] i national_thermal_plant ( ( [ \u090f\u0928\u091f\u0940\u092a\u0940\u0938\u0940] i NTPC ) ) \u0915\u094b \u092c\u0947 \u091a\u0940 \u0964 sell+past", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "(7) \u0930\u0932\u093e\u092f\u0902 \u0938 \u0928\u0947", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Reliance sold its gas to (National Thermal Power Plant) i ( NTPC i ) .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "(7) \u0930\u0932\u093e\u092f\u0902 \u0938 \u0928\u0947", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "In above example (7), \u0928\u0947 \u0936\u0928\u0932 \u0925\u092e\u0930\u094d \u0932 \u092a\u0949\u0935\u0930 \u092a\u094d\u0932\u093e\u0902 \u091f national_thermal_plant is in romanized Devanagari and there individual word's initials are taken for abbreviation. So \u090f\u0928\u091f\u0940\u092a\u0940\u0938\u0940 NT PC has a relation 'Coreference-Abbreviation' with \u0928\u0947 \u0936\u0928\u0932 \u0925\u092e\u0930\u094d \u0932 \u092a\u0949\u0935\u0930 \u092a\u094d\u0932\u093e\u0902 \u091f national_thermal_plant noun sequences.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "(7) \u0930\u0932\u093e\u092f\u0902 \u0938 \u0928\u0947", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Like Coreference-RAbbreviation relation-type, this relation-type has same significant with only directional difference between mention and its referent. As 'R' indicates the reverse direction.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-RAbbreviation :", | |
| "sec_num": "4.3.4." | |
| }, | |
| { | |
| "text": "This relation is applicable when none of the syntactic (above mentioned), semantic (mentioned in next subsection) relations between two mentions are applicable, and still they are referring to a same discourse entity. So this relation has a significance when there are no other relation are applicable between two mentions. ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Noun-Noun :", | |
| "sec_num": "4.3.5." | |
| }, | |
| { | |
| "text": "This relation 'Coreference-Noun-Verb' is applicable when from two mentions, one mention is of a noun sequence and it is referring to an event (noun-verb sequence) in discourse. In this situation, we decided to mark 'Coreference-Noun-Verb' between these two mentions.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Noun-Verb :", | |
| "sec_num": "4.3.6." | |
| }, | |
| { | |
| "text": "In this sub type of relation, we capture those relations in which, mentions are related to each other by various semantic notions. Conceptually, we can relate them through various word-net relations, though lexically they are not identical to each other but conceptually they are related to each other. These relations are mainly useful in question-answering kind of applications.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Weak Identity relations :", | |
| "sec_num": "4.4." | |
| }, | |
| { | |
| "text": "From the Hindi dependency treebank, we identified that a noun sequence and a pronoun can refer to multiple mentions. This reference type is most obvious and unique among the all where, one mention is physically or conceptually part of other referring mention or we can say current mention is a subset of referring mention.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-PartOf :", | |
| "sec_num": "4.4.1." | |
| }, | |
| { | |
| "text": "(9) [ \u0930\u093e\u091c\u0938\u094d\u0925\u093e\u0928 \u092e\u0947\u0902 ] i Rajasthan+loc \u0906\u0902 \u0927\u0940 \u0938\u0947 storm [ \u0909\u0924\u094d\u0930\u0940 \u093f\u0939\u0938\u094d\u0938\u0947 \u092e\u0947\u0902 ] i northern_part+loc \u092a\u0947 \u095c trees \u0909\u0916\u095c \u0917\u090f uproot+past \u0964 .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-PartOf :", | |
| "sec_num": "4.4.1." | |
| }, | |
| { | |
| "text": "Tree uprooted by the storm in the northern part of Rajasthan.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-PartOf :", | |
| "sec_num": "4.4.1." | |
| }, | |
| { | |
| "text": "In above example (9), mention \u0909\u0924\u094d\u0930\u0940 \u093f\u0939\u0938\u094d\u0938\u0947 northern_part is part of \u0930\u093e\u091c\u0938\u094d\u0925\u093e\u0928 Ra jasthan . It is not referring to whole entity but only some part. In this case the noun sequence \u0909\u0924\u094d\u0930\u0940 \u093f\u0939\u0938\u094d\u0938\u0947 northern_part has a 'Coreference-PartOf' relation type with \u0930\u093e\u091c\u0938\u094d\u0925\u093e\u0928 Ra jasthan .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-PartOf :", | |
| "sec_num": "4.4.1." | |
| }, | |
| { | |
| "text": "Coreference-RpartOf relation is similar to Coreference-PartOf with only directional difference.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-RPartOf :", | |
| "sec_num": "4.4.2." | |
| }, | |
| { | |
| "text": "Like Coreference-PartOf, this is also a relation between two mentions, where in mention sequence, the first mention is subset of second mention or we can say that first mention is physically or conceptually part of second mention. 10 The impact of the growing ties between India and Pakistan, is seen in the kitchen of Kashmir. People are seeing it as a result of growing relationship between two countries .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-RPartOf :", | |
| "sec_num": "4.4.2." | |
| }, | |
| { | |
| "text": "In example (10), \u0926\u094b\u0928 \u0926\u0947 \u0936 two_countries is bigger set while \u092d\u093e\u0930\u0924 India and \u092a\u093e\u093f\u0915\u0938\u094d\u0924\u093e\u0928 Pakistan are part/smaller set of it, thus \u0926\u094b\u0928 \u0926\u0947 \u0936 two_countries has a 'Coreference-RPartOf' relation with both \u092d\u093e\u0930\u0924 India and \u092a\u093e\u093f\u0915\u0938\u094d\u0924\u093e\u0928 Pakistan .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-RPartOf :", | |
| "sec_num": "4.4.2." | |
| }, | |
| { | |
| "text": "(11) [ \u0906\u092e] i Mango \u0916\u093e\u0928\u0947 \u0915\u0947 \u092c\u093e\u0926 eat+cont._after \u0909\u0938\u0928\u0947 he+NOM [ \u0917\u0941 \u0920\u0932\u0940 \u0915\u094b] i kernel+DAT \u092b\u0947\u0902 \u0915 \u093f\u0926\u092f\u093e \u0964 throw+past", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-RPartOf :", | |
| "sec_num": "4.4.2." | |
| }, | |
| { | |
| "text": "After eating mango i , he threw kernel i .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-RPartOf :", | |
| "sec_num": "4.4.2." | |
| }, | |
| { | |
| "text": "This Inferred relation thoroughly based on different lexical and semantic relation between mentions. This lexical and semantic relation (Coreference-Inferred) includes synonymous, Hyponymy and Hypernymy, Meronymy and Holonymy (Part-whole relation), Entailment, Troponymy, Antonymy. This relation occurs when an mention can be inferred or derived from its predecessor mentions. For example, \u0935\u093e\u0939\u0928 vehicle , \u0917\u093e\u095c\u0940 car both can be inferred from each other, because they share the same synset. In example (11), \u0917\u0941 \u0920\u0932\u0940 kernel has wordnet relation with \u0906\u092e mango . Thus they link each other with 'Coreference-Inferred' relation.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Inferred :", | |
| "sec_num": "4.4.3." | |
| }, | |
| { | |
| "text": "In a sentence, function or variable is a notation that specifies places, where the substitution of certain quantifiable value, may take place. This notion is related to a placeholder (a symbol that will later has some value), or a character that stands for an specified function. In above example (12), mention \u0906\u0908\u0938\u0940 -\u096e\u0966\u0969 IC\u2212803 is not at all a reference of \u093f\u0935\u092e\u093e\u0928 \u0938\u0902 \u0916\u094d\u092f\u093e Flight_number but they shared some attributes. That is, one is the function and other describes the value that the function can take. So \u0906\u0908\u0938\u0940 -\u096e\u0966\u0969 IC\u2212803 has relation 'Coreference-Function-Value' with \u093f\u0935\u092e\u093e\u0928 \u0938\u0902 \u0916\u094d\u092f\u093e Flight_number .", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference-Function-Value :", | |
| "sec_num": "4.4.4." | |
| }, | |
| { | |
| "text": "Coreference annotation is defined as a process of language corpora annotation, to indicate which textual expression have been used to co-specify the same entity in the discourse. When such an annotated corpora are collected from different coders, the reliability of the annotated data has to be quantified. Many times due to annotator's preferences, they do prejudice annotation, to standardize annotation process these types of errors have to be quantified. For coreference annotation, we used various reliability metrics to quantify the annotation scheme and annotation. As mention in (Passonneau, 2004 ), Krippendorff's alpha (Krippendorff, 2012 ) is a better metric for calculating agreement for co-reference annotation as compared to other metrics. Because it considers degree of disagreement and can be apply for more than two annotators. Similar to (Dakwale et al., 2012) as explained in (Passonneau, 2004) , we also consider coreference chain as discrete categories.", | |
| "cite_spans": [ | |
| { | |
| "start": 587, | |
| "end": 604, | |
| "text": "(Passonneau, 2004", | |
| "ref_id": "BIBREF13" | |
| }, | |
| { | |
| "start": 629, | |
| "end": 648, | |
| "text": "(Krippendorff, 2012", | |
| "ref_id": "BIBREF8" | |
| }, | |
| { | |
| "start": 856, | |
| "end": 878, | |
| "text": "(Dakwale et al., 2012)", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 895, | |
| "end": 913, | |
| "text": "(Passonneau, 2004)", | |
| "ref_id": "BIBREF13" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Inter-Annotator Study", | |
| "sec_num": "5." | |
| }, | |
| { | |
| "text": "Equation (Figure 1 ) demonstrate the Kripandorff's alpha, Where PD o is probability of observed disagreement and PD E is probability of expected disagree- Figure 1 : Kripandoff alpha ments. For r coding units and m coders, the equation calculates the agreement among the annotators by summing disagreement coefficient within and across the annotators. For every pair of values b and c (for sets), \u03b4 bc is the distance between the values. n bi is the number of times the value b occurred in ith unit. In nominal scales \u03b4 = 0 when b = c (equivalent sets); otherwise \u03b4 = 1 (different sets). The \u03b4 value in above equation is depends upon comparison of two sets. As described in (Passonneau, 2004) the relation between two sets can be describe in four different ways i.e, identity, subscription, intersection and disjunctions and their \u03b4 values are, 0 for identity, 0.33 for subsumption, .67 for intersection and 1 for disjunctions. As discussed in (Artstein and Poesio, 2008) , the assignment of \u03b4 value mislead the agreement, because it is not able to capture the length of sets (cardinality of sets). i.e. for the same sets discussed in (Passonneau, 2004) , {C, H, J, K} and {C, H} has subsumption relation and {C, H, J, K} and {C} also has subsumption relation so according to (Passonneau, 2004) , one needs to give 0.33 \u03b4 value. These two sets {C, H} and {C} have 2 and 1 elements respectively, and this difference in sets cannot be captured here. Therefore we use Jaccard index also known as the Jaccard similarity coefficient , for comparing the similarity anddiversity of sample sets, to capture this difference as mention in (Artstein and Poesio, 2008) . The Jaccard coefficient measures similarity between finite sample sets, and is defined as the size of the intersection divided by the size of the union of the sample sets (Jaccard equation - Figure 2 ). We use new \u03b4 define as follow:", | |
| "cite_spans": [ | |
| { | |
| "start": 674, | |
| "end": 692, | |
| "text": "(Passonneau, 2004)", | |
| "ref_id": "BIBREF13" | |
| }, | |
| { | |
| "start": 944, | |
| "end": 971, | |
| "text": "(Artstein and Poesio, 2008)", | |
| "ref_id": "BIBREF0" | |
| }, | |
| { | |
| "start": 1135, | |
| "end": 1153, | |
| "text": "(Passonneau, 2004)", | |
| "ref_id": "BIBREF13" | |
| }, | |
| { | |
| "start": 1276, | |
| "end": 1294, | |
| "text": "(Passonneau, 2004)", | |
| "ref_id": "BIBREF13" | |
| }, | |
| { | |
| "start": 1629, | |
| "end": 1656, | |
| "text": "(Artstein and Poesio, 2008)", | |
| "ref_id": "BIBREF0" | |
| } | |
| ], | |
| "ref_spans": [ | |
| { | |
| "start": 9, | |
| "end": 18, | |
| "text": "(Figure 1", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 155, | |
| "end": 163, | |
| "text": "Figure 1", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 1850, | |
| "end": 1858, | |
| "text": "Figure 2", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Inter-Annotator Study", | |
| "sec_num": "5." | |
| }, | |
| { | |
| "text": "EQUATION", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [ | |
| { | |
| "start": 0, | |
| "end": 8, | |
| "text": "EQUATION", | |
| "ref_id": "EQREF", | |
| "raw_str": "\u03b4 = \uf8f1 \uf8f4 \uf8f4 \uf8f2 \uf8f4 \uf8f4 \uf8f3 0, for identity 0.33 * J, for subsumption 0.67 * J, for intersection 1 * J, for dis junctions", | |
| "eq_num": "(1)" | |
| } | |
| ], | |
| "section": "Inter-Annotator Study", | |
| "sec_num": "5." | |
| }, | |
| { | |
| "text": "With the mentioned annotation scheme, we measured the inter annotator agreement score on 60 coreference annotated documents with 3 annotators on Krippendorff's alpha with Jaccard similarity index. We calculated agreement score on two different set of documents (60 each) with same annotators and got average agreement score around 87 % on 1400 co-refereeing mentions.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Inter-Annotator Study", | |
| "sec_num": "5." | |
| }, | |
| { | |
| "text": "As the Hindi dependency treebank has several layers of annotation from part-of-speech to Karaka based dependency (Begum et al., 2008) and on the top of that annotators annotated coreference relations, links and span of mentions. Annotators can refer these layered information throw annotation tool (CAT). All these detail information tuned corpora tends to lead us on relatively higher agreement score. ", | |
| "cite_spans": [ | |
| { | |
| "start": 113, | |
| "end": 133, | |
| "text": "(Begum et al., 2008)", | |
| "ref_id": "BIBREF1" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Inter-Annotator Study", | |
| "sec_num": "5." | |
| }, | |
| { | |
| "text": "We built coreference annotation tool (CAT) to help and automate the complex process of coreference annotation. This web based tool has functionality to identify annotation errors and provide an assistance to annotators. Other annotation functionality like semi-automatic annotation, textual representation are implemented after concerning with the annotators. CAT mainly deals with 3 aspects of Coreference annotation 1) Mention identification 2) Initial feed sets/chains generator 3) Automatic mention and chain head identification.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Coreference Annotation Tool (CAT)", | |
| "sec_num": "6." | |
| }, | |
| { | |
| "text": "We use automatic mention identifier tool 7 and configure it in this coreference annotation tool (CAT). From where annotators can guide them self in mention selection or directly add those mentions in further annotation process.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Mention identification :", | |
| "sec_num": "6.1." | |
| }, | |
| { | |
| "text": "On extracted mentions, CAT also provide an option to generate initial coreference chains/sets base on several string match, named dictionary and dependency relations (optional) base rules. This is up to annotators to use this facility or not because sometimes some of the rules generate noisy chains. In this kind of tedious annotation, annotators sometimes make mistakes. i.e wrong coreference chain selection for a given mention. We try to tackle these kind of issues by using derived dictionary (from dbpedia) and dependency based rules in CAT. These rules give alert to annotators on doubtable annotations. These warning messages can be ignored. Although annotators can edit tool generated automatic annotation for smooth working.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Initial feed chain generation :", | |
| "sec_num": "6.2." | |
| }, | |
| { | |
| "text": "As Hindi is a head final language (Benmamoun et al., 2009) , linguistic head of a noun sequence/mention is mostly the last word and, main verb for the event (verb-noun sequence) mention. We built automatic mention head identifier based on that observation. We also observed that for a coreference chain, its head/governing element lies in its first mention thus we also added an automatic chain head marking facility to CAT. Annotators can always edit/update/delete automated annotation to correct the annotation process.", | |
| "cite_spans": [ | |
| { | |
| "start": 34, | |
| "end": 58, | |
| "text": "(Benmamoun et al., 2009)", | |
| "ref_id": "BIBREF2" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Head Selection :", | |
| "sec_num": "6.3." | |
| }, | |
| { | |
| "text": ".1.4. Anaphora-RE :This Anaphora-RE relation is similar to Anaphora-E relation with only directional difference. Same as 'Anaphora-RC' this type of relations are also known as cataphora.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "https://github.com/vmujadia/MentionIdentifier", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [ | |
| { | |
| "text": "Around 9000 sentences of Hindi dependency treebank (Begum et al., 2008) have been annotated with coreference and their relations with our described annotation scheme. During these process CAT was used by annotators for assistance.Hindi Dep. TreeBank Size # Documents 600 # Sentences 9000 # Tokens 90000 ", | |
| "cite_spans": [ | |
| { | |
| "start": 51, | |
| "end": 71, | |
| "text": "(Begum et al., 2008)", | |
| "ref_id": "BIBREF1" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Annotated Data", | |
| "sec_num": "7." | |
| }, | |
| { | |
| "text": "In this paper, we described a scheme for coreference annotation and applied to the Hindi Dependency Treebank. In future this scheme can be validated for other languages especially Indian languages. The main contribution of this work is to define and discuss different types of co-referential relations between the referential entities to account the need of applications like question-answering, summarization, sentiment analysis, etc. The inter annotator agreement shows that proposed scheme performs consistently well.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusion and future work", | |
| "sec_num": "8." | |
| } | |
| ], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Inter-coder agreement for computational linguistics", | |
| "authors": [ | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Artstein", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Poesio", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2008, | |
| "venue": "Computational Linguistics", | |
| "volume": "34", | |
| "issue": "4", | |
| "pages": "555--596", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Artstein, R. and Poesio, M. (2008). Inter-coder agree- ment for computational linguistics. Computational Linguistics, 34(4):555-596.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "Dependency annotation scheme for indian languages", | |
| "authors": [ | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Begum", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Husain", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Dhwaj", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "M" | |
| ], | |
| "last": "Sharma", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "L", | |
| "middle": [], | |
| "last": "Bai", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Sangal", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2008, | |
| "venue": "IJCNLP", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "721--726", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Begum, R., Husain, S., Dhwaj, A., Sharma, D. M., Bai, L., and Sangal, R. (2008). Dependency annotation scheme for indian languages. In IJCNLP, pages 721- 726. Citeseer.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "Closest conjunct agreement in head final languages. Linguistic variation yearbook", | |
| "authors": [ | |
| { | |
| "first": "E", | |
| "middle": [], | |
| "last": "Benmamoun", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Bhatia", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Polinsky", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "", | |
| "volume": "9", | |
| "issue": "", | |
| "pages": "67--88", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Benmamoun, E., Bhatia, A., and Polinsky, M. (2009). Closest conjunct agreement in head final languages. Linguistic variation yearbook, 9(1):67-88.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "Ssf: Shakti standard format guide. Language Technologies Research Centre, International Institute of Information Technology", | |
| "authors": [ | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Bharati", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Sangal", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "M" | |
| ], | |
| "last": "Sharma", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2007, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "1--25", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Bharati, A., Sangal, R., and Sharma, D. M. (2007). Ssf: Shakti standard format guide. Language Tech- nologies Research Centre, International Institute of Information Technology, Hyderabad, India, pages 1- 25.", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "A framework and graphical development environment for robust nlp tools and applications", | |
| "authors": [ | |
| { | |
| "first": "H", | |
| "middle": [], | |
| "last": "Cunningham", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [], | |
| "last": "Maynard", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "K", | |
| "middle": [], | |
| "last": "Bontcheva", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "V", | |
| "middle": [], | |
| "last": "Tablan", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2002, | |
| "venue": "ACL", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "168--175", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Cunningham, H., Maynard, D., Bontcheva, K., and Tablan, V. (2002). A framework and graphical de- velopment environment for robust nlp tools and ap- plications. In ACL, pages 168-175.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "Anaphora annotation in hindi dependency treebank", | |
| "authors": [ | |
| { | |
| "first": "P", | |
| "middle": [], | |
| "last": "Dakwale", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "H", | |
| "middle": [], | |
| "last": "Sharma", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "M" | |
| ], | |
| "last": "Sharma", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "Proceedings of the 26th Pacific Asia Conference on Language, Information, and Computation", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "391--400", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Dakwale, P., Sharma, H., and Sharma, D. M. (2012). Anaphora annotation in hindi dependency treebank. In Proceedings of the 26th Pacific Asia Conference on Language, Information, and Computation, pages 391-400.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "Text and knowledge mining for coreference resolution", | |
| "authors": [ | |
| { | |
| "first": "S", | |
| "middle": [ | |
| "M" | |
| ], | |
| "last": "Harabagiu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "R", | |
| "middle": [ | |
| "C" | |
| ], | |
| "last": "Bunescu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [ | |
| "J" | |
| ], | |
| "last": "Maiorano", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2001, | |
| "venue": "Proceedings of the second meeting of the North American Chapter of the Association for Computational Linguistics on Language technologies", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "1--8", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Harabagiu, S. M., Bunescu, R. C., and Maiorano, S. J. (2001). Text and knowledge mining for coreference resolution. In Proceedings of the second meeting of the North American Chapter of the Association for Computational Linguistics on Language technolo- gies, pages 1-8. Association for Computational Lin- guistics.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "Ontonotes: the 90% solution", | |
| "authors": [ | |
| { | |
| "first": "E", | |
| "middle": [], | |
| "last": "Hovy", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Marcus", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Palmer", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "L", | |
| "middle": [], | |
| "last": "Ramshaw", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Weischedel", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2006, | |
| "venue": "Proceedings of the human language technology conference of the NAACL, Companion Volume: Short Papers", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "57--60", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Hovy, E., Marcus, M., Palmer, M., Ramshaw, L., and Weischedel, R. (2006). Ontonotes: the 90% solu- tion. In Proceedings of the human language technol- ogy conference of the NAACL, Companion Volume: Short Papers, pages 57-60. Association for Compu- tational Linguistics.", | |
| "links": null | |
| }, | |
| "BIBREF8": { | |
| "ref_id": "b8", | |
| "title": "Content analysis: An introduction to its methodology", | |
| "authors": [ | |
| { | |
| "first": "K", | |
| "middle": [], | |
| "last": "Krippendorff", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Krippendorff, K. (2012). Content analysis: An intro- duction to its methodology. Sage.", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "title": "Coreference and anaphora: developing annotating tools, annotated resources and annotation strategies", | |
| "authors": [ | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Mitkov", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Evans", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C", | |
| "middle": [], | |
| "last": "Orasan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C", | |
| "middle": [], | |
| "last": "Barbu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "L", | |
| "middle": [], | |
| "last": "Jones", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "V", | |
| "middle": [], | |
| "last": "Sotirova", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2000, | |
| "venue": "Proceedings of the Discourse, Anaphora and Reference Resolution Conference (DAARC2000)", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "49--58", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Mitkov, R., Evans, R., Orasan, C., Barbu, C., Jones, L., and Sotirova, V. (2000). Coreference and anaphora: developing annotating tools, annotated resources and annotation strategies. In Proceedings of the Discourse, Anaphora and Reference Resolu- tion Conference (DAARC2000), pages 49-58. Cite- seer.", | |
| "links": null | |
| }, | |
| "BIBREF10": { | |
| "ref_id": "b10", | |
| "title": "Annotating anaphoric and bridging relations with mmax", | |
| "authors": [ | |
| { | |
| "first": "C", | |
| "middle": [], | |
| "last": "M\u00fcller", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Strube", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2001, | |
| "venue": "Proceedings of the Second SIGdial Workshop on Dis", | |
| "volume": "16", | |
| "issue": "", | |
| "pages": "1--6", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "M\u00fcller, C. and Strube, M. (2001). Annotating anaphoric and bridging relations with mmax. In Proceedings of the Second SIGdial Workshop on Dis- course and Dialogue-Volume 16, pages 1-6. Associ- ation for Computational Linguistics.", | |
| "links": null | |
| }, | |
| "BIBREF11": { | |
| "ref_id": "b11", | |
| "title": "A coreferentially annotated corpus and anaphora resolution for czech", | |
| "authors": [ | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Nedoluzhko", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "J", | |
| "middle": [], | |
| "last": "M\u00edrovsk\u1ef3", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Nov\u00e1k", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2013, | |
| "venue": "Computational Linguistics and Intellectual Technologies. Papers from the Annual International Conference", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Nedoluzhko, A., M\u00edrovsk\u1ef3, J., and Nov\u00e1k, M. (2013). A coreferentially annotated corpus and anaphora resolution for czech. In Computational Linguistics and Intellectual Technologies. Papers from the An- nual International Conference\" Dialogue.", | |
| "links": null | |
| }, | |
| "BIBREF12": { | |
| "ref_id": "b12", | |
| "title": "Clinka a coreferential links annotator", | |
| "authors": [ | |
| { | |
| "first": "C", | |
| "middle": [], | |
| "last": "Orasan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "W", | |
| "middle": [ | |
| "W" | |
| ], | |
| "last": "Sb", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2000, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Orasan, C. and Sb, W. W. (2000). Clinka a corefer- ential links annotator.", | |
| "links": null | |
| }, | |
| "BIBREF13": { | |
| "ref_id": "b13", | |
| "title": "Computing reliability for coreference annotation", | |
| "authors": [ | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Passonneau", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2004, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Passonneau, R. (2004). Computing reliability for coreference annotation.", | |
| "links": null | |
| }, | |
| "BIBREF14": { | |
| "ref_id": "b14", | |
| "title": "Where anaphora and coreference meet. annotation in the spanish cess-ece corpus", | |
| "authors": [ | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Recasens", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [ | |
| "A" | |
| ], | |
| "last": "Marti", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Taul\u00e9", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2007, | |
| "venue": "Proceedings of RANLP. Citeseer", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Recasens, M., Marti, M. A., and Taul\u00e9, M. (2007). Where anaphora and coreference meet. annotation in the spanish cess-ece corpus. In Proceedings of RANLP. Citeseer.", | |
| "links": null | |
| }, | |
| "BIBREF15": { | |
| "ref_id": "b15", | |
| "title": "A typology of near-identity relations for coreference (nident)", | |
| "authors": [ | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Recasens", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "E", | |
| "middle": [ | |
| "H" | |
| ], | |
| "last": "Hovy", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [ | |
| "A" | |
| ], | |
| "last": "Mart\u00ed", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2010, | |
| "venue": "LREC", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Recasens, M., Hovy, E. H., and Mart\u00ed, M. A. (2010). A typology of near-identity relations for coreference (nident). In LREC.", | |
| "links": null | |
| }, | |
| "BIBREF16": { | |
| "ref_id": "b16", | |
| "title": "Bart: A modular toolkit for coreference resolution", | |
| "authors": [ | |
| { | |
| "first": "Y", | |
| "middle": [], | |
| "last": "Versley", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [ | |
| "P" | |
| ], | |
| "last": "Ponzetto", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Poesio", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "V", | |
| "middle": [], | |
| "last": "Eidelman", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Jern", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "J", | |
| "middle": [], | |
| "last": "Smith", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "X", | |
| "middle": [], | |
| "last": "Yang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Moschitti", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2008, | |
| "venue": "Proceedings of the 46th NAACL: Demo Session", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "9--12", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Versley, Y., Ponzetto, S. P., Poesio, M., Eidelman, V., Jern, A., Smith, J., Yang, X., and Moschitti, A. (2008). Bart: A modular toolkit for coreference res- olution. In Proceedings of the 46th NAACL: Demo Session, pages 9-12. Association for Computational Linguistics.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "FIGREF1": { | |
| "num": null, | |
| "text": "Figure 2: Jaccard Index", | |
| "uris": null, | |
| "type_str": "figure" | |
| }, | |
| "TABREF0": { | |
| "html": null, | |
| "text": "", | |
| "content": "<table><tr><td>Coreference annotation for Example (1)</td></tr></table>", | |
| "type_str": "table", | |
| "num": null | |
| }, | |
| "TABREF2": { | |
| "html": null, | |
| "text": "and MUC 4 [1] schemes, We can only partially derive coreference chains with some difficulties. 5. A mention cannot be a multi-chain member in both schemes ((Dakwale et al., 2012) and MUC 5 [1]. 6. In both schemes ((Dakwale et al., 2012) and MUC 6 [1] there is no provision to represent partial or full membership of a mention in chains.", | |
| "content": "<table><tr><td>From the example (1) (table -3), We can see that</td></tr><tr><td>the issues mentioned above and the ones pointed out</td></tr><tr><td>in (Dakwale et al., 2012) can resolved by our sug-</td></tr><tr><td>gested scheme. 'Markable span identification' (point</td></tr><tr><td>2.) and 'Distributed mention(Dakwale et al., 2012)'</td></tr><tr><td>issues can be solved by annotating coreference on</td></tr></table>", | |
| "type_str": "table", | |
| "num": null | |
| }, | |
| "TABREF3": { | |
| "html": null, | |
| "text": "For example, \u092e\u0941 \u0902 \u092c\u0908 \u092a\u0941 \u0932\u0938 mumbai\u2212police and \u092a\u0941 \u0932\u0938 police , it is clear that \u092e\u0941 \u0902 \u092c\u0908 \u092a\u0941 \u0932\u0938 mumbai\u2212police is not referring to whole \u092a\u0941 \u0932\u0938 police but refers to sub-part of it. This deep taxonomy of relation is quite useful for question-answering and summarization alike tasks. Inferred and function-value relations are also part of weak identity relations. Following subsections explain all coreference relation types one-by-one in detail.", | |
| "content": "<table><tr><td colspan=\"2\">(2) [ \u092d\u093e\u091c\u092a\u093e \u0928\u0947 \u0924\u093e] i</td><td>[ \u0905\u092a\u0928\u0947 ] i</td><td>\u0909\u0917\u0930\u094d</td><td>\u0924\u0947 \u0935\u0930</td><td>\u092e\u0947\u0902</td></tr><tr><td colspan=\"2\">BJP_leader</td><td>his.POSS</td><td>fiery</td><td>temper</td><td>in</td></tr><tr><td>\u0915\u0941 \u091b</td><td colspan=\"3\">\u0928\u0939\u0940\u0902 \u0915\u0939\u0928\u093e \u091a\u093e\u0939\u0924\u0947 \u0925\u0947 \u0964</td><td/></tr><tr><td>anything</td><td colspan=\"4\">not_want_to_say+Past</td></tr><tr><td colspan=\"6\">BJP leader did not want to say anything in his</td></tr><tr><td colspan=\"2\">fiery temper.</td><td/><td/><td/></tr></table>", | |
| "type_str": "table", | |
| "num": null | |
| }, | |
| "TABREF8": { | |
| "html": null, | |
| "text": "\u092a\u0930\u094d\u092c\u0902 \u0927\u0915 manager is referring to second mention \u0938\u0941 \u092c\u094b\u0926\u0940\u092a \u091a\u0915\u0930\u094d\u0935\u0924\u0940\u0930\u094d Subodeep_Chakraborty . There is no syntactic linkage between mention \u092a\u0930\u094d\u092c\u0902 \u0927\u0915 manager and mention \u0938\u0941 \u092c\u094b\u0926\u0940\u092a \u091a\u0915\u0930\u094d\u0935\u0924\u0940\u0930\u094d Subodeep_Chakraborty thus, mention \u092a\u0930\u094d\u092c\u0902 \u0927\u0915 manager has 'Coreference-Noun-Noun' relation with mention \u0938\u0941 \u092c\u094b-\u0926\u0940\u092a \u091a\u0915\u0930\u094d\u0935\u0924\u0940\u0930\u094d Subodeep_Chakraborty .", | |
| "content": "<table><tr><td>(8) \u0918\u091f\u0928\u093e \u0915 \u0938\u0942 \u091a\u0928\u093e</td><td colspan=\"3\">[ \u092a\u0930\u094d\u092c\u0902 \u0927\u0915 ] i</td><td>[</td></tr><tr><td colspan=\"2\">Incident's_information'</td><td colspan=\"2\">manager</td></tr><tr><td colspan=\"2\">\u0938\u0941 \u092c\u094b\u0926\u0940\u092a \u091a\u0915\u0930\u094d\u0935\u0924\u0940\u0930\u094d \u0915\u094b] i</td><td/><td>\u0926\u0940 \u0964</td></tr><tr><td colspan=\"3\">Subodeep_Chakraborty.DAT</td><td>give+past</td></tr><tr><td>[ \u092a\u0930\u094d\u092c\u0902 \u0927\u0915 \u0928\u0947 ] i</td><td>\u092a\u0941 \u0932\u0938 \u0915\u094b</td><td colspan=\"2\">\u0907\u0938 \u0918\u091f\u0928\u093e \u0938\u0947</td></tr><tr><td>Manager</td><td>police.DAT</td><td colspan=\"2\">this_incident</td></tr><tr><td>\u0905\u0935\u0917\u0924 \u0915\u0930\u093e\u092f\u093e \u0964</td><td/><td/></tr><tr><td>inform+past</td><td/><td/></tr><tr><td colspan=\"4\">Information about incident was given to</td></tr><tr><td colspan=\"4\">Manager i (Subodeep Chakraborty) i . Manager i</td></tr><tr><td colspan=\"4\">informed police about the incident.</td></tr><tr><td colspan=\"4\">In above example (8), three mentions (with i in-</td></tr><tr><td colspan=\"4\">dex) are in referring to same entity, where last men-</td></tr><tr><td>tion</td><td/><td/></tr></table>", | |
| "type_str": "table", | |
| "num": null | |
| }, | |
| "TABREF10": { | |
| "html": null, | |
| "text": "\u0938\u0902 \u0916\u094d\u092f\u093e number , \u0924\u093e\u092a\u092e\u093e\u0928 temperature , \u0915\u0930\u094d\u092e\u093e\u0902 \u0915", | |
| "content": "<table><tr><td/><td/><td/><td>It</td></tr><tr><td colspan=\"4\">is also somewhere related to copula, but here we are</td></tr><tr><td colspan=\"4\">accounting only those functions which are capable to</td></tr><tr><td colspan=\"4\">holding only quantifiable values. Here for example, it</td></tr><tr><td>can be anything like rank and \u0917\u0941 \u0923 mark , etc.</td><td/><td/><td/></tr><tr><td>(12) [ \u093f\u0935\u092e\u093e\u0928 \u0938\u0902 \u0916\u094d\u092f\u093e] i</td><td/><td colspan=\"2\">[ \u0906\u0908\u0938\u0940 -\u096e\u0966\u0969 \u0915\u0947 ] i</td></tr><tr><td colspan=\"2\">Flight_number</td><td>IC-803's</td><td/></tr><tr><td>\u0915\u0948 \u092a\u094d\u091f\u0928 \u0928\u0947</td><td colspan=\"2\">\u093f\u092c\u0928\u093e \u0938\u092e\u092f \u0917\u0902 \u0935\u093e\u090f</td><td/></tr><tr><td>caption.ACC</td><td colspan=\"3\">without_time_weasting</td></tr><tr><td>\u093f\u0935\u092e\u093e\u0928 \u0915\u094b</td><td>\u0935\u093e\u092a\u0938</td><td>\u093f\u0926 \u0940 \u0915 \u0924\u0930\u092b</td><td>\u092e\u094b\u095c \u093f\u0926\u092f\u093e \u0964</td></tr><tr><td>aircraft.DAT</td><td>back</td><td>to_Delhi.LOC</td><td>turn+past</td></tr><tr><td colspan=\"4\">(Flight Number) i (IC -503) i 's' caption turned</td></tr><tr><td colspan=\"4\">the aircraft back to Delhi without losing time .</td></tr></table>", | |
| "type_str": "table", | |
| "num": null | |
| } | |
| } | |
| } | |
| } |