| { | |
| "paper_id": "W87-0105", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T06:44:36.335950Z" | |
| }, | |
| "title": "", | |
| "authors": [], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "", | |
| "pdf_parse": { | |
| "paper_id": "W87-0105", | |
| "_pdf_hash": "", | |
| "abstract": [], | |
| "body_text": [ | |
| { | |
| "text": "The arguments are kind of typed variables. The expression could also be stated: produce(X,Y) and country(X) and vessel(Y ). An argument may further be another predication. If the word 'countries' were replaced by 'nordic countries' then the resulting structure would be: ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "VESSEL PL", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "PRODUCE j +-------f*--- SITUATION I 4-----+----+ I I", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "VESSEL PL", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "I -------^-------------I- I X4", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "4------------^-------------4-", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "The node with \"niiden\" (plural and genetive of \"se\") is replaced with reference to the structure bound to variable X3.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "4------------^-------------4-", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Often an e l l i p t i c a l sentence is preceeded by a complete sentence, which contains the le x ica l en tities le ft out from the e l lip t ic a l sentence.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "ELLIPSIS", | |
| "sec_num": "3.2" | |
| }, | |
| { | |
| "text": "The f i r s t problem is to v erify whether an expression is e l li p t i c a l or not. -a certain expression is used together with e l lip t ic a l utterance. ( \"Enta.s . . .\" , \"What about . . .\") -the case of nominal phrase is other than nominative.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "ELLIPSIS", | |
| "sec_num": "3.2" | |
| }, | |
| { | |
| "text": "(\"Annen?\", \"Ann's?\") -nominal phrase is in comparative ( \"Enemman kuin HelsingissM.?\", \"More than in Helsinki?\") -tran sitive verb has no object ( \"Myyty vuonna 1983?\",\"Sold year 1983?\") 74 Proceedings of NODALIDA 1987 ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "ELLIPSIS", | |
| "sec_num": "3.2" | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Luettele yritykset, jo i l la on y ii 1000 ty o n te k ija i!", | |
| "authors": [], | |
| "year": null, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "\"Luettele yritykset, jo i l la on y ii 1000 ty o n te k ija i!\" (Give a lis t of the companies that have over 1000 employees!)", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "FIGREF0": { | |
| "uris": null, | |
| "text": "--------^-------\u00ce I ROOT-ONLY XI X2 The second example is an instance of a rule for covering expressions stating an entity to have a certain property. The entity here could be a company, companies, companies in certain branch etc. The expression is allowed to have an sp ecifica tion of point or interval of time. X2 (AKO: attribute) ANYORDER ANYOPTS (REST (AKO: time_expression)) --------+----------I --h-COMPANY PL ------^I BRANCH forestry -------^--------------\u00ce I", | |
| "type_str": "figure", | |
| "num": null | |
| }, | |
| "FIGREF1": { | |
| "uris": null, | |
| "text": "\"Mill\u00e4 y rity k sill\u00e4 on y li 1000 ty\u00f6ntekij\u00e4\u00e4?\" (Which companies have over 1000 employees) \"Mik\u00e4 on niiden vuosittainen liikevaihto?\" (What is their annual turnover?) --liik e v a ih to ------------j.-------------", | |
| "type_str": "figure", | |
| "num": null | |
| }, | |
| "FIGREF2": { | |
| "uris": null, | |
| "text": "expression were (2) \"Is the number of employees of Huhtamaki greater than the average in forestry\" it would most propahly be understood as e l li p t i c .Our tranformation rule example for analysis of anaphora had the discourse history as set of logico-semantic forms. AWARE-system may also be used to match parts of current expression towards", | |
| "type_str": "figure", | |
| "num": null | |
| }, | |
| "TABREF0": { | |
| "num": null, | |
| "text": "PREnDICATION GRAPHS AS CANONICAL REPRESENTATION OF QUERY SENTENCES Honkela, T . , Lehtola, A. and Valkonen, K.", | |
| "content": "<table><tr><td/><td/><td/><td colspan=\"7\">KIELIKONE-proiect, SITRA Foundation</td></tr><tr><td/><td/><td/><td/><td colspan=\"6\">P.O.Box 329, SF-00121 Helsinki</td></tr><tr><td/><td/><td/><td/><td/><td/><td/><td colspan=\"3\">Finland</td></tr><tr><td/><td/><td/><td/><td/><td colspan=\"5\">tel. inti + 358 0 641 877</td></tr><tr><td colspan=\"10\">This paper surveys problems encountered in</td><td>studying</td><td>the</td><td>logico-semantic</td></tr><tr><td colspan=\"4\">form and discourse</td><td colspan=\"3\">problems</td><td>of</td><td colspan=\"2\">Finnish query sentences.</td><td>W e ca ll the</td></tr><tr><td colspan=\"10\">logico-semantic form a predication graph.</td><td>The basic framework we use</td><td>to</td></tr><tr><td>represent</td><td>the</td><td colspan=\"3\">logica l</td><td colspan=\"3\">form of</td><td colspan=\"2\">Finnish query sentences is an annotated</td></tr><tr><td colspan=\"5\">logical tree transformed</td><td colspan=\"2\">from</td><td>the</td><td colspan=\"2\">dependency</td><td>parse</td><td>tree</td><td>using</td><td>graph</td></tr><tr><td colspan=\"2\">transformations</td><td>of</td><td colspan=\"6\">the AWARE-system.</td><td>Examples of analysing e l l i p t i c and</td></tr><tr><td>anaphoric</td><td colspan=\"3\">expressions</td><td/><td>are</td><td colspan=\"2\">given.</td><td/><td>Finally,</td><td>some</td><td>c r it ic a l</td><td>points</td><td>of</td></tr><tr><td colspan=\"9\">computational semantics are discussed.</td></tr><tr><td>Keywords:</td><td colspan=\"9\">semantics, predication, discourse analysis</td></tr><tr><td colspan=\"4\">1 CANONICAL STRUCTURE</td><td/><td/><td/><td/><td/></tr><tr><td colspan=\"10\">One of the main objectives of the logico-semantic level is to</td><td>give</td><td>those</td></tr><tr><td>utterances</td><td colspan=\"9\">that d iffe r only syntactically a uniform representation.</td><td>This</td></tr><tr><td colspan=\"3\">representation is</td><td colspan=\"5\">a predication</td><td colspan=\"2\">structure</td><td>with</td><td>predicates</td><td>and</td><td>their</td></tr><tr><td>arguments.</td><td colspan=\"2\">For</td><td colspan=\"3\">example,</td><td>the</td><td colspan=\"3\">expressions</td><td>\"v a ltio t,</td><td>iotka</td><td>tuottavat</td></tr><tr><td colspan=\"10\">laivoja\" (countries that produce vessels) and \"laivola tuottavat</td><td>v a ltiot\"</td></tr><tr><td>(countries</td><td colspan=\"3\">producing</td><td colspan=\"3\">vessels)</td><td colspan=\"2\">would</td><td>lead into a single logico-semantic</td></tr><tr><td colspan=\"7\">form something like as follow s:</td><td/><td/></tr><tr><td/><td colspan=\"2\">PRODUCE</td><td/><td/><td/><td/><td/><td/></tr><tr><td>C O U NT RY</td><td/><td/><td/><td/><td/><td/><td/><td/></tr><tr><td>PL</td><td/><td/><td/><td/><td/><td/><td/><td/></tr></table>", | |
| "html": null, | |
| "type_str": "table" | |
| }, | |
| "TABREF1": { | |
| "num": null, | |
| "text": "The fir s t rule instance matches to dependency structures (for", | |
| "content": "<table><tr><td/><td/><td/><td/><td/><td/><td/><td colspan=\"3\">-70 -</td></tr><tr><td/><td/><td/><td/><td/><td/><td/><td/><td/><td>more</td><td>about</td></tr><tr><td colspan=\"10\">Thi3 could be also expressed with a clause: dependency parsing see Lehtola et alw l9 8 5 or Valkonen et a l..l9 8 7 ) where produce(X,Y) and country'X;</td></tr><tr><td colspan=\"10\">and the dependent is restricted to be a particular branch and situ a tion (X .n ord ic) and vessel(Y ). Our representational form makes the regent is</td></tr><tr><td colspan=\"10\">the variables in the logical form im plicit thus making it any synonym for company. This very simple rule contains only semantic more readable.</td></tr><tr><td colspan=\"10\">The conditions in addition to the choise of the variable to be passed to an upper predication is demand dependency structure sp ecified. Further</td></tr><tr><td colspan=\"10\">driven. syntactic checks could be added to avoid overgeneration. The solution is based on the ideas of polymorphism.</td></tr><tr><td colspan=\"5\">2 CHOICE OF PREDICATES</td><td/><td/><td/><td/></tr><tr><td colspan=\"8\">There are two essential decisions to</td><td colspan=\"2\">be</td><td>made when the</td><td>predicates</td><td>for</td></tr><tr><td colspan=\"8\">logico-semantic form are selected.</td><td colspan=\"2\">F irstly , does the system interpret the</td></tr><tr><td colspan=\"10\">semantic content of utterances strongly thus making possibly</td><td>also</td><td>strong</td></tr><tr><td colspan=\"2\">reduction</td><td colspan=\"8\">or does the system rely on the original form by using the verbs</td></tr><tr><td colspan=\"7\">with their valences as predicates</td><td colspan=\"2\">with</td><td>arguments?</td><td>Secondly,</td><td>one</td><td>must</td></tr><tr><td colspan=\"10\">decide whether the predicates are general, sp e cific or both.</td></tr><tr><td>2.1</td><td colspan=\"9\">GENERAL PREDICATIONS IN NLI FOR DATABASES</td></tr><tr><td colspan=\"10\">The selection of predications is not determined by</td><td>the</td><td>features</td><td>of</td><td>the</td></tr><tr><td colspan=\"3\">AWARE-system.</td><td colspan=\"2\">General</td><td colspan=\"3\">predicates</td><td colspan=\"2\">as well as sp e cific ones may be used.</td></tr><tr><td colspan=\"10\">The degree of canonization depends on the person(s) who makes the semantic</td></tr><tr><td colspan=\"2\">modeling,</td><td>too.</td><td>In</td><td colspan=\"6\">the natural language interface for databases, SUOM EX</td></tr><tr><td colspan=\"3\">(JAppinen & a l.</td><td colspan=\"6\">1988) we use general</td><td>predicates.</td><td>At</td><td>conceptual</td><td>level</td></tr><tr><td>these</td><td colspan=\"9\">predicates re fle ct the en tity-a ttrib u te-rela tion sh ip (EAR) approach</td></tr><tr><td colspan=\"5\">for conceptual modeling.</td><td colspan=\"2\">This</td><td>is</td><td colspan=\"2\">motivated</td><td>by</td><td>the</td><td>fact</td><td>that</td><td>these</td></tr><tr><td colspan=\"10\">predicates r e fle c t the conceptual models of the databases.</td></tr><tr><td colspan=\"10\">Let's assume that we have companies with a branch of business and certain</td></tr><tr><td colspan=\"2\">properties</td><td>(or</td><td colspan=\"3\">a ttrib u tes).</td><td colspan=\"4\">Here we have</td><td>an example of using AW ARE</td></tr><tr><td colspan=\"10\">transformation rules to create predication structure for expressions</td><td>like</td></tr><tr><td>\"Anna</td><td colspan=\"5\">metScLalan yhtididen</td><td colspan=\"4\">liik e v a ih to I\"</td><td>(Give</td><td>the</td><td>turnover</td><td>of</td><td>the</td></tr><tr><td colspan=\"4\">companies in fo re stry ).</td><td/><td/><td/><td/><td/></tr><tr><td colspan=\"4\">Proceedings of NODALIDA 1987</td><td/><td/><td/><td/><td/><td>70</td></tr></table>", | |
| "html": null, | |
| "type_str": "table" | |
| }, | |
| "TABREF2": { | |
| "num": null, | |
| "text": "To emphasize the generality of AWARE, one must state that", | |
| "content": "<table><tr><td/><td/><td/><td/><td/><td/><td/><td>the</td><td>choice</td><td>of</td></tr><tr><td colspan=\"2\">predicates</td><td colspan=\"3\">such as</td><td colspan=\"3\">' HAVE-PROPERTY'</td><td>follows from its use as a part of a</td></tr><tr><td colspan=\"4\">database in te rfa ce .</td><td/><td/><td/></tr><tr><td>2.2</td><td colspan=\"5\">CONCEPTUAL HIERARCHY</td><td/></tr><tr><td colspan=\"7\">Nodes in graph transformation rules</td><td>may contain</td><td>semantic</td><td>restrictions.</td></tr><tr><td>For</td><td colspan=\"7\">each re strictio n a proper level of generality is needed.</td><td>Information</td></tr><tr><td colspan=\"8\">about conceptual classes is in a form of hierarchy (compare</td><td>to</td><td>Grozs</td><td>et</td></tr><tr><td>a l.</td><td>1987).</td><td colspan=\"2\">The</td><td>use</td><td colspan=\"3\">of semantic restriction s and their relation to the</td></tr><tr><td colspan=\"8\">conceptual hierarchy could be exemplified with pair</td><td>of</td><td>expressions</td><td>like</td></tr><tr><td>(1)</td><td colspan=\"2\">\"Peter's</td><td colspan=\"5\">car\" versus (2) \"Peter's w ife\".</td><td>The f i r s t expression could</td></tr><tr><td colspan=\"8\">be transformed into predication 'OWN' but the latter one</td><td>presumably</td><td>not.</td></tr><tr><td>The</td><td colspan=\"3\">c la s s ific a tio n</td><td colspan=\"2\">into</td><td>living</td><td>and non-living</td><td>objects can be used to</td></tr><tr><td colspan=\"8\">refine the transformation to match appropriately.</td></tr><tr><td colspan=\"8\">3 D ISCOURSE ANALYSIS WITH GRAPH TRANSFORMATIONS</td></tr><tr><td colspan=\"8\">In many cases it is not possible to interpret a sentence</td><td>without</td><td>solving</td></tr><tr><td colspan=\"2\">references</td><td>to</td><td>the</td><td colspan=\"2\">other</td><td colspan=\"2\">sentences</td><td>of the discourse.</td><td>The AWARE-system</td></tr><tr><td colspan=\"8\">makes it possible to analyze also the context of an utterance rather</td><td>than</td></tr><tr><td>only</td><td colspan=\"7\">a single dependency structure.</td><td>The expressions of the discourse are</td></tr><tr><td colspan=\"8\">gathered under a single node called 'Discourse Node' (DN).</td></tr><tr><td/><td colspan=\"4\">Discourse_node</td><td/><td/></tr><tr><td/><td/><td>I</td><td/><td/><td/><td/></tr></table>", | |
| "html": null, | |
| "type_str": "table" | |
| } | |
| } | |
| } | |
| } |