| { | |
| "paper_id": "W89-0104", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T03:44:45.958150Z" | |
| }, | |
| "title": "P\u00e5 vej mod en fagsproglig tekstfortolker?", | |
| "authors": [ | |
| { | |
| "first": "Steffen", | |
| "middle": [ | |
| "Leo" | |
| ], | |
| "last": "Hansen", | |
| "suffix": "", | |
| "affiliation": {}, | |
| "email": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Bstract", | |
| "suffix": "", | |
| "affiliation": {}, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "The FAGFLADE Project \"FAGFLADE\" is short for Danish \"fagsproglig graenseflade\" (\"special purpose language interface\"). The aim of the FAG FLADE project is to de velop and test theories and methods for automatic interpretation o f texts written in special purpose language. We use the expression \"text inter preter\" to designate a program which transfers the information contained in a natural language text into knowledge representations in a knowledge base. Thus a text interpreter is meant to perform part o f the task o f a knowledge acquisition system in an expert system. The FAG FLADE project takes as its basis the text of the Danish Com panies A ct (\"Lov om aktieselskaber\"). It is not our ambition to build an expert system, nor do we aim at the construction of a complete text inter pretation system, fit for use, and capable o f translating statutes into legal knowledge brises. Rather, we take the development of a specific interpreter to be an ideal goal which defines an overall project capable of giving rise to a handful o f interesting subprojects for the investigation o f general theories and principles concerning interpreters, e.g. in the domain of syntactic and semantic analysis, parsing strategies, knowledge representation, dictionary databases, and terminological analysis. The choice o f the Danish Companies Act as a text basis is motivated partly by our desire to work with an LSP central to industry and com merce, pcirtly by the evident practical perspectives which will open up if text interpreters for legal directives become a realistic possibility. This paper deals with the initial phase of the project. In this phase we have constructed a framework for the syntactic and semantic analysis of simple sentence structures, and we have developed a program, written in Quintus Prolog, which carries out syntactic and semantic parsing of these structures.", | |
| "pdf_parse": { | |
| "paper_id": "W89-0104", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "The FAGFLADE Project \"FAGFLADE\" is short for Danish \"fagsproglig graenseflade\" (\"special purpose language interface\"). The aim of the FAG FLADE project is to de velop and test theories and methods for automatic interpretation o f texts written in special purpose language. We use the expression \"text inter preter\" to designate a program which transfers the information contained in a natural language text into knowledge representations in a knowledge base. Thus a text interpreter is meant to perform part o f the task o f a knowledge acquisition system in an expert system. The FAG FLADE project takes as its basis the text of the Danish Com panies A ct (\"Lov om aktieselskaber\"). It is not our ambition to build an expert system, nor do we aim at the construction of a complete text inter pretation system, fit for use, and capable o f translating statutes into legal knowledge brises. Rather, we take the development of a specific interpreter to be an ideal goal which defines an overall project capable of giving rise to a handful o f interesting subprojects for the investigation o f general theories and principles concerning interpreters, e.g. in the domain of syntactic and semantic analysis, parsing strategies, knowledge representation, dictionary databases, and terminological analysis. The choice o f the Danish Companies Act as a text basis is motivated partly by our desire to work with an LSP central to industry and com merce, pcirtly by the evident practical perspectives which will open up if text interpreters for legal directives become a realistic possibility. This paper deals with the initial phase of the project. In this phase we have constructed a framework for the syntactic and semantic analysis of simple sentence structures, and we have developed a program, written in Quintus Prolog, which carries out syntactic and semantic parsing of these structures.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "r\u00e5 d e r k n y tte t til u d v ik lin g e n a f en te k s tfo rto lk e r, d e t v il f\u00f8 rst o g frem m est sige g ru n d la g e t fo r d en sy n ta k tis k e o g sem a n tisk e a n a ly se, v id e n sre p rae se n ta tio n , ud v ik lin g o g a fp r \u00f8 v n in g a f p a rsin g stra te g ie r, e t k o n s titu e n ts tru k tu rtrae , d els en sem a n tisk stru k tu r. D e n sy n ta k tisk e stru k tu r a n v e n d e s ikke i \u00f8 v r ig t a f te k stfo rto lk e re n , h v o r im o d d en sem a n tisk e stru k tu r skal d a n n e u d g a n g s p u n k t fo r d e n v id e n sre p rae se n ta tio n s o m skal ov erf\u00f8res til v id e n s b a s e n . (1 ) s (n p (d e t (d e n n e ), n (a ft a le )), v p (v (p \u00e5 f\u00f8 r e r ), n p (n (s e ls k a b e t )), n p (d e t (e n ), n (fo r p lig t e ls e ))))", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduktion", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "(2 ) p \u00e5 f\u00f8 r e (a g e n t(a fta le ), th e m e (fo r p lig te ls e ), lo c u s (s e ls k a b )) G r u n d la g e t fo r d en sy n ta k tisk e a n a ly se o g den sem an tiske rep rae sen ta tion ( rep rae se n tere r p \u00e5 d en en e sid e en v a len sg ra m m a tisk n y tae n k n in g , o g er p \u00e5 den a n d e n sid e d e n fu n k tio n so m ru m m er n \u00f8 g len til ra tion a lise rin g en i an tallet a f fu n k tio n e r. r e en sem a n tisk re p rae se n ta tio n a f d e s im p le sae tn in ger v i a rb e jd e r m ed . N \u00e5 r v i skal i g a n g m e d a t e ta b le r e s elv e v id e n s b asen kan d e t im id le rtid m eg et vel v is e s ig a t d e n n e sem a n tisk e re p rae se n ta tio n er u tilstrae k k elig , o g a t fx . a n tallet a f ro lle r skal su p p leres. X' Y*Z*pftf0re(agent(Z).theme(Y).locus(X)) undtager v_np_pp sub(agent) obj(theme) adj(locus) X*Y~Z*undtage(agent(Z).theme(Y).locus(X)) U n d er pa rsn in gen a f sae tn in g en \"D e n n e a fta le p \u00e5 f\u00f8 re r selsk a b et en fo r p lig te ls e \" vil parseren -n \u00e5r den skal b y g g e d en sem a n tisk e stru k tu r fo r sae tn in g e n - ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduktion", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "1 ) S -> NP VP (2 ) VP -> V (3 ) VP -> V NP (4 ) VP -> V PP (5 ) VP -> V AP (6 ) VP -> V NP NP (7 ) VP -> V NP PP (8 ) VP -> V NP AP (9 ) VP -> V PP PP (10 ) PP -> P NP (11 ) PP -> P AP (12 ) NP -> N (1 3 ) NP -> DET N (1 4 ) AP -> A (1 5 ) AP -> A PP I lek sik on er h v", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduktion", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "Proceedings of NODALIDA 1989", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Computational Linguistics -Reykjavik 1989a g e n t svaxer til d en tra d itio n e lle o p fa tte ls e a f ag en t o g kan kun a ssocieres m ed", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "FAGFLADE: Text, Types and Tokens, Lambda", | |
| "authors": [ | |
| { | |
| "first": "Litteratur", | |
| "middle": [], | |
| "last": "Hansen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Steffen", | |
| "middle": [], | |
| "last": "Leo", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": null, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Litteratur Hansen, Steffen Leo (forthcoming), FAGFLADE: Text, Types and Tokens, Lambda.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "FAGFLADE: The Initicil Phcise of a Project in Natural Language Interpretation. Lambda, 11. Institut for Datalingvistik", | |
| "authors": [ | |
| { | |
| "first": "Steffen", | |
| "middle": [], | |
| "last": "Hansen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Carl", | |
| "middle": [], | |
| "last": "Leo", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Vikner", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1989, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Hansen, Steffen Leo and Carl Vikner. 1989. FAGFLADE: The Initicil Phcise of a Project in Natural Language Interpretation. Lambda, 11. Institut for Datalingvistik, HHk.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "On Valence and Grammatical Relations", | |
| "authors": [ | |
| { | |
| "first": "Michael", | |
| "middle": [], | |
| "last": "Herslund", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1988, | |
| "venue": "Copenhagen Studies on Language", | |
| "volume": "11", | |
| "issue": "", | |
| "pages": "3--35", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Herslund, Michael. 1988a. On Valence and Grammatical Relations. Copenhagen Studies on Language, 11:3-35, Copenhagen.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "Le datif en frangais", | |
| "authors": [ | |
| { | |
| "first": "Michael", | |
| "middle": [], | |
| "last": "Herslund", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1988, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Herslund, Michael. 1988b. Le datif en frangais. Louvain-Paris, Editions Peeters.", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "En valensgrammatisk fremstilling. I. Verbernes syntaks. Romansk Institut", | |
| "authors": [ | |
| { | |
| "first": "Michael", | |
| "middle": [], | |
| "last": "Herslund", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Finn", | |
| "middle": [], | |
| "last": "S\u00f8rensen", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1985, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Herslund, Michael and Finn S\u00f8rensen. 1985. De franske verber. En valensgrammatisk fremstilling. I. Verbernes syntaks. Romansk Institut, K\u00f8benhavns Universitet.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "DICTED: An Editor for Dictionaries Stored as Prolog Data batses", | |
| "authors": [ | |
| { | |
| "first": "Henrik", | |
| "middle": [], | |
| "last": "Kersting", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1989, | |
| "venue": "Lambda", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Kersting, Henrik. 1989. DICTED: An Editor for Dictionaries Stored as Prolog Data batses. Lambda, 12.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "PC-Grammar: An Alternative?", | |
| "authors": [ | |
| { | |
| "first": "Hanne", | |
| "middle": [], | |
| "last": "Korzen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Henning", | |
| "middle": [], | |
| "last": "N\u00f8lke", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Henrik", | |
| "middle": [], | |
| "last": "Prebensen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Finn", | |
| "middle": [], | |
| "last": "S\u00f8rensen", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1983, | |
| "venue": "Acta Linguistica Hafniensia", | |
| "volume": "18", | |
| "issue": "", | |
| "pages": "5--53", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Korzen, Hanne, Henning N\u00f8lke, Henrik Prebensen and Finn S\u00f8rensen. 1983. PC- Grammar: An Alternative?, Acta Linguistica Hafniensia, 18:5-53, Copenhagen.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "Om rollen Locus, Notat. Institut for Datalingvistik, Handelsh\u00f8j skolen i K\u00f8benhavn", | |
| "authors": [ | |
| { | |
| "first": "Finn", | |
| "middle": [], | |
| "last": "S\u00f8rensen", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1988, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "S\u00f8rensen, Finn. 1988. Om rollen Locus, Notat. Institut for Datalingvistik, Handelsh\u00f8j skolen i K\u00f8benhavn.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "FIGREF0": { | |
| "type_str": "figure", | |
| "text": "P \u00e5 In stitu t fo r D a ta lin g v is tik p \u00e5 H a n d e lsh \u00f8 jsk o le n i K \u00f8 b e n h a v n h ar v i siden in stitu ttet b le v o p r e tte t i 1985 u d v ik le t en tr a d itio n fo r et fo rs k n in g s s a m a rb e jd e . 41 Proceedings of NODALIDA 1989, pages 41-50 Computational Linguistics -Reykjavik 1989 d er i staxten fo r m e d e sig s o m en rae kk e d a ta lin g v istisk e sem in arer o v e r u d v a lg te em n e r, m en s o m sid e n 1987 h a r f\u00f8 r t til a r b e jd e t p \u00e5 et e g en tlig t fo rsk n in g sp ro je k t m e d d e lta g e ls e a f a lle fo rs k n in g s m e d a rb e jd e re , n em lig p r o je k te t F A G F L A D E , h v ilk et st\u00e5 r fo r '/\u00ab^s p r o g lig g iae n s e fla d e '. M \u00e5 le t fo r d e tte p r o je k t er a t a fp r \u00f8 v e o g u d v ik le te orie r o g m e to d e r b a g k o n s tru k tio n e n a f en te k stfo rto lk e r, d v s . et sy stem so m a n a ly serer en tekst o g o v e r f\u00f8 r e r v id e n h e rfra til v id e n sre p rae se n ta tio n i en v id e n sb a se . E n s\u00e5 d a n tek st fo r to lk e r , fo re s tille r v i o s, kan in d g \u00e5 s o m k o m p o n e n t i e t e k sp e rtsy ste m som en d el a f et v id e n s in d lae rin g s m o d u l. D e n tek st v i h a r v a lg t s o m g ru n d la g fo r p r o je k te t er d en da n sk e ,JjOv o m a k tie s e ls k a b e r \", o g d e t s p r o g s o m er rep rae sen teret i lo v en er d e rm e d det su b la n g u a g e s o m fo rto lk e re n skal k u n n e a n a ly ser o g forst\u00e5 . P r o je k t e t sig te r ikke m o d a t u d v ik le et fae rd ig t e k sp e rtsy ste m , m en er ud g a n g s p u n k t fo r en rae k ke m in d re p r o je k te r d e r fok u serer p \u00e5 p r o b le m e r o g o m", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF1": { | |
| "type_str": "figure", | |
| "text": "d en m ask in lae sb a re o r d b o g s stru k tu r o g o m fa n g sa m t te rm in o lo g is k e p r o b le m e r s \u00e5 s o m id en tifik ation o g reprae senta^ tio n a f fle rle d ed e te rm e r, s y n o n y m e r , d e fin itio n e r o g b e g reb srela tion er. I d a g e r p r o je k te t n \u00e5 et d e rtil a t d e r er u d v ik le t en parser i Q u in tu s P r o lo g s o m kan k la re sim p le sae tn in g s stru k tu rer, d v s. sae tn in ger der kun in d eh o ld er v a le n sb u n d n e led s o m o p tr ae d e r p \u00e5 d en fo r v e n te d e p o s itio n i sae tn in gen . V i kan a lts \u00e5 ik k e p \u00e5 n u v ae re n d e tid s p u n k t k la re sp \u00f8 rg e b isae tn in g e r eller top ikaliserin ger, h eller ik ke \u00a3idverbialer, le d sae tn in g e r o g a d sk illig e a n d re tin g. D e t o u t p u t v i f\u00e5 r fra p a rseren e r d els sae tn in gen s s y n t2dctiske s tru k tu r i fo r m a f", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF2": { | |
| "type_str": "figure", | |
| "text": "E t in p u t so m D e n n e a fta le p \u00e5 f\u00f8 re r selska b et en fo rp lig te lse v il d e r fo r g iv e fig. o u tp u t:", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF3": { | |
| "type_str": "figure", | |
| "text": "er d els en k o n s titu e n ts tru k tu r g ra m m a tik , dels e t lek sik on m e d en leksikalsk b e sk riv e lse a f v e r b e r n e u d fr a en v a len sb a seret k o m b in a tio n a f g ra m m a tisk e fu n k tio n e r o g roller. J e g skal i d e t f\u00f8 lg e n d e k o m m e n ae rm ere in d p \u00e5 d e ov erv ejelser d er lig g e r til g ru n d fo r d e n n e leksikalske b esk riv else o g p \u00e5 d en s im p le m en te rin g i s e lv e p a rs e rp ro g ra m m e t.", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF4": { | |
| "type_str": "figure", | |
| "text": "v i h ar b y g g e t e r en ' le ft-c o r n e r , b o t t o m -u p ' p arser. D e n sta rte r s\u00e5ledes m ed d e t f\u00f8 r s te o r d i en in p u t-sek v en s o g fo r s \u00f8 g e r a t o p b y g g e en k onstitu en tstru k tu r fo r et S s o m h a r d e tte f\u00f8 rste o r d s o m sit v e n stre h j\u00f8 r n e . G r a m m atikk en d er ru m m er flg. g en sk riv n in g sreg le r, h v o r a f 2 -9 rep rae sen terer d e sim p le sae tn in g sstru k tu rer s o m p a rsern e kan klare n e to p nu:", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF5": { | |
| "type_str": "figure", | |
| "text": "ert en k elt o r d fo r m fo r s y n e t m e d en o p ly s n in g o m k on stitu te n tty p e . S u b k a teg oriserin g en a f V P 'ern e g e n fin d es i lek sik on s o m en su bk ateg oriserin g sra m m e, d e r rep rae sen terer d e k o n s titu te n tty p e r s o m v e r b e t kan fo r b in d e s m e d s o m v alen sled. D e n n e k o n s titu e n tra m m e skal u n ificeres m e d en id en tisk ra m m e i en P r o lo g reg el o g tje n e r d e r fo r d e t fo r m \u00e5 l a t e ffek tiv ise re selv e p a rsn in g en . H v is u n ifica tion en lyk k es, v il p a rseren v ae lg e d en p \u00e5 g ae ld e n d e regel. V i kan s\u00e5 led es b e g y n d e a t se p \u00e5 o p b y g n in g e n a f en p o s t i lek sik on d e r h a r fo r m a f en P r o lo g k lau su l m ed p rae d ik a tet d / 6 h vis a rg u m e n te r a lle er s a m m e n sa tte term er: d(word(_), lexeme(_), lexcat(_), g r a m _ f o n n ( _ ) , a y n t . s p e c s e m _ s p e c F or v e r b e t ' p \u00e5 f\u00f8 r e ' s o m o p tr ae d e r i e k se m p e ls ae tn in g en o v e n fo r ser u d fy ld n in gen a f d e f\u00f8 rste a rg u m e n te r s\u00e5 led es u d : u m e n te t lex ca t in d e h o ld e r o p ly s n in g o m k on stitu e n tty p e , s y n t s p e c hax t o sa m m e n s a tte term er s o m a rg u m en te r: d els c o n sL fra m e, d er a n g iv er k o n stitu te n tr a m m e n , d els ad j (a d je k t ), d e r so m a rg u m en t h a r en liste ov er tilla d te p r ae p o s itio n e r hvis d er kan o p tr ae d e et v a len sb u n d e t P P . N o ta tio n e n [ n i l ] an g iv e r a t d e t ikke kan fo r e k o m m e v e d d e tte v erb u m . 3 Funktioner D e fu n k tio n e r d e r o p e re re s m ed i F A G F L A D E er h en tet fr a d en v alen steori s o m H erslu n d o g S \u00f8ren sen (1 9 8 5 , 1988a, 1 9 8 8 b ) har u d v ik le t i deres a rb e jd e m e d fran sk e v e r b e r . D e r kan h er k un b liv e p la d s til at sk itsere h o v ed trae k k en e i d en n e te o r i, d e r b y g g e r p \u00e5 en d istin k tio n m e lle m sae tn in gsled d er h \u00f8rer til hele sae tn in g e n , frie tilf\u00f8 je ls e r, o g sae tn in g s led s o m er d ik tere t a f d e t k o n k rete verb u m s o m o p tr ae d e r i sae tn in g en . E t va len sled d efin eres so m det sy n ta k tisk e u d try k for d e re la tio n e r d e r b e s t\u00e5 r m ellem et v e rb u m s o m sem an tisk p rae d ik a t o g d ets argu m e n te r o g m ellem a rg u m e n te r in d b y rd e s , o g d e t er d e rfo r kun verbaJafh ae n gige sae tn in g s le d m ed a rg u m e n tsta tu s d er kan o p fa tte s so m valen sled. E n d v id e r e an ta g e s d e t a t v a len sled a len e kan b esk riv e s i fo r m a f fu n k tio n e r, o g at d er til en h ver lek sikalsk b e sk riv else a f et v e r b u m h \u00f8 re r et valen ssk em a m ed a n g iv else a f h vilk e fu n k tio n e r , resp . v a le n sled , d e t p \u00e5 g ae ld e n d e v e rb u m fo rb in d e s m ed . D e r o p e re re s m e d et in v en ta r p \u00e5 tre g ra m m a tisk e fu n k tio n e r: s u b je k t, o b je k t o g a d je k t. H ertil h a r v i i F A G F L A D E fo r eg en re g n in g f\u00f8 je t en fje rd e fu n k tio n , a g en tia l, fo r o g s \u00e5 a t k u n n e h \u00e5 n d te re p a ssiv .S a m m e n lig n e t m e d a n ta llet a f valen sled i tra d itio n e lle valen sg ra m m a tik k er er d e r sket en d ra stisk r e d u k tio n i a n ta llet a f fu n k tio n e r. F u n k tion en a d jek t", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF6": { | |
| "type_str": "figure", | |
| "text": "T r a d itio n e lt er m a n van t til at a r b e jd e m ed va len sled so m D O (d a t iv o b je k t ), l O (in d ire k te o b je k t ) , A eller L O (a d v e r b ia l/lo k a t iv o b je k t ), S P o g O P (h h v . s u b je k ts -o g o b je k ts p r ae d ik a t). V ed a n v en d elsen a f et g iv e t v e rb u m kan m an s\u00e5 led es k o m m e ud fo r a t d e t er m u ligt a t v ae lg e m ellem en k o n stru k tion h v or v e r b e t o p tr ae d e r m ed D O eller 1 0 : m an kan fx . ' g iv e n o g en n o g e t' eller ' giv e n o g et til n o g e n ', eller d e t kan v ae re et va lg m ellem D O o g L O : 'je g har efte rla d t h a m m in e b \u00f8 g e r ' o v e r fo r 'je g h a r e fte rla d t m in e b \u00f8 g e r h os h a m ' . H vad m an d e r im o d ikke kan k o m m e u d fo r , er en k o m b in a tio n a f b e g g e fu n k tion er sa m tid ig , d v s.at d e o v e n n ae v n te , tr a d itio n e lle v alen sled g e n sid ig t u delu k ker h in a n den . D e rfo r a n ta g es d e t, a t d e alle rep rae sen terer en o g sa m m e led fu n k tio n so m im id lertid kan realiseres p \u00e5 forsk e llig m \u00e5 d e. V i kan illu strere det m ed flg. ek sem p ler:(1 ) H a n e fte r lo d b \u00f8 r n e n e en b e tr a g te lig form u eD O O(2 ) H a n e fte r lo d en b e tr a g te lig fo rm u e til b \u00f8 rn e n eO 1 0 (3 ) H an e fte rlo d sin b o g h os sin b r o d e r O L O (4 ) H an e fte r lo d sin b il p \u00e5 b a n eg \u00e5 rd en O L O (5 ) H an e fte r lo d h e n d e d y p t fru streret O P A r g u m e n ta tio n e n fo r a t sa m m e n fa tte i d e tte tilfae ld e D O , 1 0 , L O o g O P i een o g sa m m e fu n k tio n , adjekt, er, a t u an set h vilken b e te g n e lse m a n v ae lg e r at an v en d e s\u00e5 g ae ld e r d e t i alle k o n stru k tio n e r a t fu n k tio n e n s o m s\u00e5 d a n e ta b le r e r en n y re la tio n i sae tn in ge n , i o v e n n ae v n te tilfae ld e en re la tion m elle m o b jek t + adjekt. D e n n e re la tio n o p fa tte s so m en sek u n d ae r p rsed ik ation d e r kan p 2irafraseres m ed e n ten V AE R E eller H A V E o g so m su p p le re r v e r b e ts b e ty d n in g v e d a t p la ce re referen ten fo r O i fo r h o ld til referen ten fo r a d je k te t. P la cerin g en kan v ae re en ten m eg et k o n k ret s o m i e k se m p lern e (3 ) : b o g e n er h os h ans b r o d e r , o g (4 ): bilen er p \u00e5 b a n e g \u00e5 rd e n , eller m ere a b stra k t s o m i (1 ): b \u00f8 rn e n e h ar en b e tr a g te lig form u e. D e r kan im id le rtid o g s \u00e5 v ae r e ta le o m en p la ce rin g i fo r h o ld til en eg en sk a b eller en klasse, n e m lig i d e tilfae ld e h v or a d jek tet svarer til en p rae d ik a tiv k o n stru k tio n m e d e n ten S P eller, so m i (5 ), m ed O P . B egru n d elsen fo r a t in d f\u00f8 r e a d je k t er d e r fo r p \u00e5 d en en e sid e a t d e t er a fh ae n g ig t a f sae tn in gen s v e rb a l so m a rg u m e n t i lo g isk fo r s ta n d , o g p \u00e5 d e n a n d en sid e at det so m a rg u m en t g \u00f8 r d et m u lig t a t rela tere et a f d e t o fu n d a m e n ta le va len sled, S o g O , til et a n d e t a rg u m en t i sae tn in g en o g a t d e n n e rela tion m e llem va len sled in d b y rd e s er u d try k fo r en sek u n d ae r p rae d ik a tio n . U d fy ld n in g e n a f a d je k te t kan v ae re forsk ellig , m en d en sek u n d ae re p rae d ik a tio n er k o n sta n t. S om tid lig ere n ae v n t o m fa tte r d en leksikalske b esk riv else a f et v e rb u m et valen ssk em a so m a n g iv er h vilke fu n k tio n e r d e r kan fo r e k o m m e sa m m en m e d et givet v e rb u m . E t s\u00e5 d a n t v a len ssk em a fin d es o g s \u00e5 i d e t lek sik on s o m v o re s p a rser b e n y tte r, m en in d en v i skal se n ae rm ere p \u00e5 d e tte v il d e t v ae r e n \u00f8 d v e n d ig t at b e h a n d le d e tem a tisk e roller, id et d e g ra m m a tisk e fu n k tio n e r i va len ssk em a et associeres m ed en sem a n tisk rolle. 4 Roller M ed u d g a n g sp u n k t i o p fa tte ls e n a f v e r b e t s o m e t p rae d ik a t m e d tilh \u00f8 r e n d e ar g u m en ter kan d e s em a n tisk e r e la tio n e r m ellem a rg u m e n te rn e b e sk riv es v h ja . sem an tiske roller. D en sem a n tisk e a n a ly se s t \u00f8 t te r sig lig eled es p \u00e5 H erslu n d o g S \u00f8ren sen (1 9 8 5 ) sa m t e n d v id e re p \u00e5 K o r z e n et alii (1 9 8 3 ) o g d eres a r b e jd e m e d d en s\u00e5k aldte P K -g r a m m a tik , id e t d en o v e n fo r o m t a lte v a len steo ri o g d e sy n ta k tiske fu n k tion e r er u d g a n g sp u n k t fo r a t a sso cie re fu n k tio n e r m e d roller. A n ta lle t a f roller sy n e s a t v ae re en te m m e lig u b e s te m m e lig s t\u00f8rre lse, m en v i h ar foretru k k et a t red u cere a n ta llet i h e n h o ld til d e n ae v n te a r b e jd e r o g i v o re s p r o je k t in d sk rae n k et d e t til tre: a g en t, te m a o g lo cu s. fu n k tio n e rn e su b jek t o g agential. l o c u s o p tr ae d e r ku n v e d p rae d ik a te r m e d m in d st t o a rgu m en ter. D en n e rolle re la te rer referen ten fo r et amdet a rg u m e n t til d en lo c a tio n som lo cu sa rg u m e n te t u d try k k er. L o cu s kan a ssocie res m ed fu n k tion ern e su bjekt, adje k t o g agential. t e m a er sem a n tisk set d en m este g en erelle o g d en m in d st p rae cise a f d e tre roller. R e fe re n te n fo r te m a er d en ' e n tit y ' s o m en ten p \u00e5 v irk es a f h a n d lin gen eller p la ce res m h t. en lo c a t io n . T em a kan a ssocieres m ed fu n k tion ern e su bjet, o b je c t o g adjekt. D is tr ib u tio n o g k o m b in a tio n a f ro lle r a sso cie re t m ed fu n k tio n e r kan s\u00e5ledes se u d s o m v ist i flg. sk em a et: em a et fre m g \u00e5 r a t fx . v e r b e t a n se o p tr ae d e r m e d en fu n k tion sra m m e ' s u b je k t + o b je k t + a d je k t ' o g a t d e r d e rtil svarer ro llera m m e n ' ag en t + te m a + lo c u s ' . U d tr y k k e t ' n o n e ' b e ty d e r a t d en p \u00e5 g ae ld e n d e fu n k tio n ikke kan fo re k om m e v e d d e t a n f\u00f8 r te v e rb u m . J eg v il g e rn e k n y tte en en k elt k o m m e n ta r til rollen lo c u s m ed udgcin gspun kt i d isse sae tn in g er: (1 ) sp \u00f8 rg s m \u00e5 le t b o rtfa ld er s \u00f8 d v e n d ig e a f b e sty re lse n a g t ( a g e n t ) I s ae tn in g (4 ) o g (5 ) e r lo c u s a sso cie re t m ed fu n k tio n e n a d je k t, s o m h er svarer til h h v . d e t tra d itio n e lle s u b je k ts p r ae d ik a tiv (4 ) o g o b je k ts p r ae d ik a tiv (5 ), o g i (6 ) e r lo c u s a sso cie re t m e d et a d je k t d e r svarer til d e t tra d itio n e lle d a tiv o b je k t. J eg a n f\u00f8 r te tid lig e re a t rollen lo c u s kun o p tr ae d e r sa m m en m ed et an d et v a len sled o g at d en d e n o te r e r en lo c a t io n fo r referen ten d e n o te re t a f d et an d et valen sled. S er v i p \u00e5 d e tre sae tn in g e r o v e n fo r s\u00e5 er d e t s u b je k te t , d isse aktier, d e r i (4 ) loca liseres ek ste n sio n a lt i fo r h o ld til en klasse, k lassen a f fo n d s a k tie r , a d je ktet, o g i (5 ) er d e t o b je k t e t , d isse u n d ers\u00f8 g els er, d e r lo ca lise re s in te n sio n a lt m e d h en syn til eg en sk a b en a t v ae r e n \u00f8 d v e n d ig , u d tr y k t i a d je k te t. L ig e s o m fo r fu n k tion ern es v e d k o m m e n d e kan lo c a t io n e n o g s \u00e5 v ae r e m e re a b s t r a c t s\u00e5 d a n s o m d e t er tilfae ld et i (6 ) h v o r o b je k t e t , en fo rp lig telse, lo ca lise re s i fo r h o ld til selskabet, a d je k te t, fo rst\u00e5 e t p \u00e5 d e n m \u00e5 d e a t d et er selsk a b et s o m hax en fo rp lig te lse . V i kan h e rm ed a fslu tte u d fy ld n in g e n a f a rg u m e n te r i le k s ik o n p o s te n o g in d sae tte v a len ssk em aet i fo r m a f d e sem a n tisk e sp e cifik a tion er: il ikke k o m m e y d erlig e re in d p \u00e5 rollern e h er. P \u00e5 n u v ae re n d e tid sp u n k t i p r o je k te t fu n gerer d en n e o p fa tte ls e a f ro ller k o m b in e re t m e d v a len steo rien u d m ae rk et n \u00e5r det g ae ld er o m a t p r o d u c e", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF7": { | |
| "type_str": "figure", | |
| "text": "e o p ly sn in g e r d e r lig g e r i lek sik on so m v ist o v e n fo r in d ta ste s m a n u elt v e d h jae lp a f en sae rlig e d ito r , D I C T E D , u d v ik le t a f H en rik K e r s tin g (u n d e r u d g iv e lse ) i fo rb in d else m e d F A G F L A D E p r o je k te t . D e d a n n er u d g a n g s p u n k t fo r en p r o c e d u re s o m gen ererer et n yt lek sik on i fo r m a f en p r o lo g d a ta b a s e , h v o r le x e m n a v n e t a ssocieres m ed en sem a n tisk re p rae se n ta tio n i fo r m a f e t la m b d a u d tr y k b e re g n e t p \u00e5 b a g g ru n d a f a rg u m en tern e i d en sa m m e n sa tte te rm f u n c ( t i o n ) _ t o _ r o l e . D e t skal h er in d sk y d es, a t ku n v e rb e r o g a d je k tiv e r, h e ru n d er o g s \u00e5 p a r tic ip ie r , p \u00e5 n u v ae ren d e tid sp u n k t fo re k o m m e r m e d en sp e cifik se m a n tisk re p rae se n ta tio n . A lle a n d re o rd k la sser, d v s . d e te rm in a tiv e r, p rae p o s itio n e r o g s u b s ta n tiv e r, h a r d et p \u00e5 g ae ld e n d e lek sem n a v n s o m sem a n tisk rep rae se n ta tio n . S a m m e n s a tte N P 'er h ar Computational Linguistics -Reykjavik 1989 k e rn ele d d ets lek sem n a v n s o m sem a n tisk rep rae sen ta tio n o g P P 'er h ar styrelsens kern e s o m se m a n tisk re p rae se n ta tio n . U d try k k e n e aktie, a k tiern e, disse a k tier o g a f d isse a k tie r h a r s\u00e5 le d es a lle en o g sa m m e rep rae sen ta tio n , n em lig aktie. A r g u m e n te r n e i te rm e n f u n c _ t o _ r o l e er d e fire sy n ta k tisk e fu n k tion er i o r d n e t rae k k ef\u00f8lg e: s u b j, o b j , a d j, a gl, a fb ild e t p \u00e5 d e sem a n tisk e roller a gen t, tem a o g lo c u s . A lle fire fu n k tio n e r o p tr ae d e r , o g s \u00e5 i d e tilfae ld e h v o r d e rent fak tisk ik k e kan fo r e k o m m e r o g d e r fo r so m v ae r d i h ar ' n o n e ' so m v i s\u00e5 d et ov en for. D e a n v e n d es v e d k o n stru k tio n e n a f la m b d a u d tr y k k k e t, hvis a rg u m en ter ligeledes fo r e k o m m e r i k a n on isk rae k k ef\u00f8lg e: a g e n t ( _ ) , th em e ( _ ) , l o c u s ( _ ) . V e rb e t ' p \u00e5 f\u00f8 r e ' h ar s\u00e5 led e s fig. f u n c _ t o _ r o l e rep rae sen ta tion : f u n c _ t o _ r o l e ( s u b ( a g e n t ) , o b j ( t h e m e ) , a d j ( l o c u s ) , a g l ( n o n e ) ) s o m k on v e rtere s til fig . la m b d a u d tr y k : X~Y \" Z~p \u00e5 f \u00f8 r e ( a g e n t ( Z ) , theme(Y), locus(X) ) D e t t e u d tr y k b ereg n es a f fig. regel: V, Ver b + adject + object (V NP NP) . \u2022/. Verb + object + adject (V NP PP & V NP AP) . b u i l d _ L a m b d a \u00a3 x p r ( L e x e m e . f u n c _ t o _ r ole(sub(RoleS),obj(RoleO),adj(RoleAd),a g l ( n o n e ) ) , X\"Y~Z\"Expr) :-Rol e S \\== none, RoleO \\== none, R o l e A d \\== none, RoleS T e r m =.. [RoleS,Z ] , RoleO T e r m =.. [RoleO,Y], RoleA d T e r m =.. [RoleAd,X], s o r t . R o l e T e r m s ( [RoleSTerm,R o l e O T e r m , R o l e A d T e r m ] , S o r t e d R T ) , Exp r =.. [Lexeme|SortedRT]. N \u00e5 r p a rseren skal b y g g e en sae tn in g s sem a n tisk e stru k tu r har d en d a for v e rb e rn e s v e d k o m m e n d e d en ty p e in fo r m a tio n e r til sin r\u00e5 d ig h ed so m v ist p \u00e5 fig. sid e.", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF8": { | |
| "type_str": "figure", | |
| "text": "K o lo n n e n m e d fu n k tio n e r o g ro lle r kan lae ses p \u00e5 to m \u00e5 d er: d els s o m u d try k fo r en a fb ild n in g a f syntaJetiske fu n k tio n e r p \u00e5 roller, dels kan m an fo r h v ert ver b u m a flae se en fu n k tio n s r a m m e o g en ro lle ra m m e v e d a len e a t lae se fu n k tion e r o g roller. R a m m e r n e u d try k k e r tilla d te k on tek ster fo r et v e rb u m i fo r m a f kons titu e n te r, fu n k tio n e r eller roller. S tru k tu rern e, den sy n ta k tisk e o g sem an tiske, b y g g e s a f p a rseren v e d a t re la tere o r d e n e i d en k on k re te sae tn in g til k on stitute n tty p e r , ro lle r eller fu n k tio n e r.", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF9": { | |
| "type_str": "figure", | |
| "text": "sta rte m e d a t id e n tificere d en n e aftale s o m et N P d e r o p b e v a r e s i en v a ria b el. I leksik on u n d er p \u00e5 f\u00f8 r e r v il den f\u00e5 o p ly s n in g e r o m d en k o n s titu e n tra m m e d er g ae ld er fo r v e r b e t sa m t d e t la m b d a u d tr y k d er svarer til v e r b e t. U d fr a en m a tc h in g a f k on stitu en tra m m e n fin d er d en frem til d en reg el s o m skal a n v en d e s o g iden tificerer h ere fter en k o n stitu te n tstru k tu r m ed t o N P 'e r efte r v e r b e t. D e t f\u00f8 rste N P , selskabet, a ssocieres m ed rollen lo c u s s o m er tild e lt a d je k te t, d e t a n det N P , e n fo rp lig telse, m ed rollen te m a so m er tild e lt o b je k t e t . T il sid st v il d en u d fy ld e rollen a g en t m ed s u b je k te t , d et f\u00f8 rste N P d e n lae ste , o g a flev ere d e t fae rd ig e resu ltat: p\u00e5f\u00f8re(ag e n t ( a f t a l e ) , t h e m e (forpligtelse), l o c u s (selskab)) D en sem an tisk e re p rae se n ta tion v i kan f\u00e5 fre m nu er b a se re t p \u00e5 ro lle stru k tu re n fo r d i v i anser den fo r et cen tra lt elem en t n \u00e5 r d e n in fo r m a tio n sae tn in g en in d e h o ld e r skal fa sth o ld es s o m v id e n sre p rae se n ta tio n i v id e n sb a se n . V i er g o d t klar o v e r at d er er flere forsk ellig e sem a n tisk e fae n o m e n e r s o m ikke er rep rae sen teret i d en n u v ae ren d e fase, tid sre la tio n e r, m o d a lite t eller fx . a d v e rb ia lse m a n tik . V i har o g s \u00e5 en liste ov e r p u n k te r so m skal l\u00f8ses o g a n d re p u n k te r s o m fo re s t\u00e5 r. D e t", | |
| "uris": null, | |
| "num": null | |
| } | |
| } | |
| } | |
| } |