Stvaranje i ekstrakcija metafora korištenjem velikih jezičnih modela: Perspektive računalnih formalizacija kognitivno-jezikoslovnih fenomena
Uloženo v:
| Název: | Stvaranje i ekstrakcija metafora korištenjem velikih jezičnih modela: Perspektive računalnih formalizacija kognitivno-jezikoslovnih fenomena |
|---|---|
| Autoři: | Perak, Benedikt |
| Informace o vydavateli: | 2025. |
| Rok vydání: | 2025 |
| Témata: | Metaphor Generation, Large Language Models, Pragmatic Implications, GPT-4, generiranje metafora, gpt-4, veliki jezični modeli, kognitivna lingvistika, pragmatične implikacije, Cognitive Linguistics |
| Popis: | The paper examines the capabilities of large language models in understanding and generating metaphorical expressions. The research focuses on two key approaches to creating metaphorical expressions: 1) based on the merging of target and source conceptual domains from the Metanet.hr database, evaluating the conventionality, quality of mapping, and sentiment of the created metaphorical expressions, and 2) profiling the target domain according to arbitrarily defined pragmatic implications. Methodologically, it presents processes for shaping instructions and patterns for creating API queries, automation for self-analysis. The results are critically evaluated, including limitations in following specific instructions or creating biased or inaccurate metaphorical linguistic patterns. As accompanying resources, the paper includes tabular representations of created metaphorical expressions and a Python script for automation and evaluation of the process. The analysis indicates the GPT-4 model can synthesize and map different semantic domains, enabling the creation of appropriate metaphorical expressions. Additionally, there is an open question of metaphorical mapping based on specific pragmatic implications, where preliminary research suggests a high degree of adaptability of the model. The paper confirms the great potential of large language models in the analysis and creation of metaphors, while also highlighting the epistemological limitations of their use and the necessity for a deeper understanding of the role of language models in cognitive linguistics research, computational formalization of linguistic phenomena, and linguistic creativity. Rad razmatra sposobnosti velikih jezičnih modela pri razumijevanju i stvaranju metaforičkih izraza. Istraživanje se usredotočuje na dva ključna pristupa stvaranja metaforičkih izraza; 1) temeljem spajanja ciljnih i izvornih konceptualnih domena iz Metanet.hr baze uz vrednovanje konvencionalnosti, kvalitete preslikavanja i sentimenta stvorenih metaforičkih izraza, te 2) profiliranje ciljne domene prema proizvoljno zadanim pragmatičnim implikacijama. Metodološki se predstavljaju procesi oblikovanja instrukcija i obrazaca za stvaranje API upita, automatizaciju (samo-)analize. Kritički se procjenjuju rezultati, uključujući ograničenja u slijeđenju specifičnih uputa ili stvaranje pristranih ili netočnih metaforičnih jezičnih obrazaca. Kao pripadajući resursi, radu su priloženi tablični prikazi stvorenih metaforičkih izraza te Python skripta za automatizaciju i procjenu procesa. Analiza ukazuje da model GPT-4 posjeduje sposobnost sinteze i mapiranja različitih semantičkih domena, omogućavajući stvaranje prikladnih metaforičkih izraza. Dodatno je otvoreno pitanje metaforičnog mapiranja na temelju specifičnih pragmatičnih implikacija, gdje preliminarna istraživanja sugeriraju visok stupanj adaptabilnosti modela. Rad potvrđuje velike mogućnosti velikih jezičnih modela u analizi i stvaranju metafora, istovremeno ističući epistemološka ograničenja upotrebe i nužnosti dubljeg razumijevanja uloge jezičnih modela pri istraživanjima u kognitivnoj lingvistici, računalnoj formalizaciji jezičnih fenomena te jezične kreativnosti. |
| Druh dokumentu: | Part of book or chapter of book |
| Přístupové číslo: | edsair.dris...01492..a69531d43152f74998d1ca1198ca2e2b |
| Databáze: | OpenAIRE |
Nájsť tento článok vo Web of Science | FullText | Text: Availability: 0 CustomLinks: – Url: https://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=EBSCO&SrcAuth=EBSCO&DestApp=WOS&ServiceName=TransferToWoS&DestLinkType=GeneralSearchSummary&Func=Links&author=Perak%20B Name: ISI Category: fullText Text: Nájsť tento článok vo Web of Science Icon: https://imagesrvr.epnet.com/ls/20docs.gif MouseOverText: Nájsť tento článok vo Web of Science |
|---|---|
| Header | DbId: edsair DbLabel: OpenAIRE An: edsair.dris...01492..a69531d43152f74998d1ca1198ca2e2b RelevancyScore: 964 AccessLevel: 3 PubType: Book PubTypeId: book PreciseRelevancyScore: 963.988342285156 |
| IllustrationInfo | |
| Items | – Name: Title Label: Title Group: Ti Data: Stvaranje i ekstrakcija metafora korištenjem velikih jezičnih modela: Perspektive računalnih formalizacija kognitivno-jezikoslovnih fenomena – Name: Author Label: Authors Group: Au Data: <searchLink fieldCode="AR" term="%22Perak%2C+Benedikt%22">Perak, Benedikt</searchLink> – Name: Publisher Label: Publisher Information Group: PubInfo Data: 2025. – Name: DatePubCY Label: Publication Year Group: Date Data: 2025 – Name: Subject Label: Subject Terms Group: Su Data: <searchLink fieldCode="DE" term="%22Metaphor+Generation%22">Metaphor Generation</searchLink><br /><searchLink fieldCode="DE" term="%22Large+Language+Models%22">Large Language Models</searchLink><br /><searchLink fieldCode="DE" term="%22Pragmatic+Implications%22">Pragmatic Implications</searchLink><br /><searchLink fieldCode="DE" term="%22GPT-4%22">GPT-4</searchLink><br /><searchLink fieldCode="DE" term="%22generiranje+metafora%22">generiranje metafora</searchLink><br /><searchLink fieldCode="DE" term="%22gpt-4%22">gpt-4</searchLink><br /><searchLink fieldCode="DE" term="%22veliki+jezični+modeli%22">veliki jezični modeli</searchLink><br /><searchLink fieldCode="DE" term="%22kognitivna+lingvistika%22">kognitivna lingvistika</searchLink><br /><searchLink fieldCode="DE" term="%22pragmatične+implikacije%22">pragmatične implikacije</searchLink><br /><searchLink fieldCode="DE" term="%22Cognitive+Linguistics%22">Cognitive Linguistics</searchLink> – Name: Abstract Label: Description Group: Ab Data: The paper examines the capabilities of large language models in understanding and generating metaphorical expressions. The research focuses on two key approaches to creating metaphorical expressions: 1) based on the merging of target and source conceptual domains from the Metanet.hr database, evaluating the conventionality, quality of mapping, and sentiment of the created metaphorical expressions, and 2) profiling the target domain according to arbitrarily defined pragmatic implications. Methodologically, it presents processes for shaping instructions and patterns for creating API queries, automation for self-analysis. The results are critically evaluated, including limitations in following specific instructions or creating biased or inaccurate metaphorical linguistic patterns. As accompanying resources, the paper includes tabular representations of created metaphorical expressions and a Python script for automation and evaluation of the process. The analysis indicates the GPT-4 model can synthesize and map different semantic domains, enabling the creation of appropriate metaphorical expressions. Additionally, there is an open question of metaphorical mapping based on specific pragmatic implications, where preliminary research suggests a high degree of adaptability of the model. The paper confirms the great potential of large language models in the analysis and creation of metaphors, while also highlighting the epistemological limitations of their use and the necessity for a deeper understanding of the role of language models in cognitive linguistics research, computational formalization of linguistic phenomena, and linguistic creativity.<br />Rad razmatra sposobnosti velikih jezičnih modela pri razumijevanju i stvaranju metaforičkih izraza. Istraživanje se usredotočuje na dva ključna pristupa stvaranja metaforičkih izraza; 1) temeljem spajanja ciljnih i izvornih konceptualnih domena iz Metanet.hr baze uz vrednovanje konvencionalnosti, kvalitete preslikavanja i sentimenta stvorenih metaforičkih izraza, te 2) profiliranje ciljne domene prema proizvoljno zadanim pragmatičnim implikacijama. Metodološki se predstavljaju procesi oblikovanja instrukcija i obrazaca za stvaranje API upita, automatizaciju (samo-)analize. Kritički se procjenjuju rezultati, uključujući ograničenja u slijeđenju specifičnih uputa ili stvaranje pristranih ili netočnih metaforičnih jezičnih obrazaca. Kao pripadajući resursi, radu su priloženi tablični prikazi stvorenih metaforičkih izraza te Python skripta za automatizaciju i procjenu procesa. Analiza ukazuje da model GPT-4 posjeduje sposobnost sinteze i mapiranja različitih semantičkih domena, omogućavajući stvaranje prikladnih metaforičkih izraza. Dodatno je otvoreno pitanje metaforičnog mapiranja na temelju specifičnih pragmatičnih implikacija, gdje preliminarna istraživanja sugeriraju visok stupanj adaptabilnosti modela. Rad potvrđuje velike mogućnosti velikih jezičnih modela u analizi i stvaranju metafora, istovremeno ističući epistemološka ograničenja upotrebe i nužnosti dubljeg razumijevanja uloge jezičnih modela pri istraživanjima u kognitivnoj lingvistici, računalnoj formalizaciji jezičnih fenomena te jezične kreativnosti. – Name: TypeDocument Label: Document Type Group: TypDoc Data: Part of book or chapter of book – Name: AN Label: Accession Number Group: ID Data: edsair.dris...01492..a69531d43152f74998d1ca1198ca2e2b |
| PLink | https://erproxy.cvtisr.sk/sfx/access?url=https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=edsair&AN=edsair.dris...01492..a69531d43152f74998d1ca1198ca2e2b |
| RecordInfo | BibRecord: BibEntity: Languages: – Text: Undetermined Subjects: – SubjectFull: Metaphor Generation Type: general – SubjectFull: Large Language Models Type: general – SubjectFull: Pragmatic Implications Type: general – SubjectFull: GPT-4 Type: general – SubjectFull: generiranje metafora Type: general – SubjectFull: gpt-4 Type: general – SubjectFull: veliki jezični modeli Type: general – SubjectFull: kognitivna lingvistika Type: general – SubjectFull: pragmatične implikacije Type: general – SubjectFull: Cognitive Linguistics Type: general Titles: – TitleFull: Stvaranje i ekstrakcija metafora korištenjem velikih jezičnih modela: Perspektive računalnih formalizacija kognitivno-jezikoslovnih fenomena Type: main BibRelationships: HasContributorRelationships: – PersonEntity: Name: NameFull: Perak, Benedikt IsPartOfRelationships: – BibEntity: Dates: – D: 01 M: 01 Type: published Y: 2025 Identifiers: – Type: issn-locals Value: edsair |
| ResultId | 1 |