Napredna pretraga

Pregled bibliografske jedinice broj: 958570

Corpus-based Approach to the Extraction of the Emotional Concepts and Their Ontological Relations Using the Natural Language Logic Operators


Perak, Benedikt; Ban Kirigin, Tajana
Corpus-based Approach to the Extraction of the Emotional Concepts and Their Ontological Relations Using the Natural Language Logic Operators // Book of Abstracts
Dubrovnik, Hrvatska, 2018. str. 57-58 (predavanje, međunarodna recenzija, sažetak, znanstveni)


Naslov
Corpus-based Approach to the Extraction of the Emotional Concepts and Their Ontological Relations Using the Natural Language Logic Operators

Autori
Perak, Benedikt ; Ban Kirigin, Tajana

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
Book of Abstracts / - , 2018, 57-58

Skup
Logic and Applications 2018

Mjesto i datum
Dubrovnik, Hrvatska, 24-28.9.2018

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Ontology, natural language, logical connectives, Emotional Concepts

Sažetak
The paper deals with the identification, extraction and cross-linguistic comparison of the emotion concepts and their relationship with other material, psychological and socio-cultural concepts. What are the salient semantic domains and conceptual structures of the linguistic construals that are used to express emotions as entities (nouns), processes (verbs) and properties (adjectives, adverbs) in the communication? The methodology of this ontological corpus-based study includes three phases. The first phase deals with the construction of the Ontological Model of Concepts and Linguistic Constructions database that aims to formalize the meta-data about the ontological features of the psychological concepts and their relation with material and socio- cultural concepts. The ontological model is theoretically grounded in the system theory (Emmeche et al. 1997, Baas & Emmeche 1997, El-Hani & Emmeche 2000, Searle 2006, Capra & Luisi 2014), and cognitive approaches to the categorization (Rosch 2005). The ontological model is stored in a graph property database Neo4j (https://neo4j.com/). The second phase includes the extraction of the nominal, adjectival and processual lexical concepts related to the psychological phenomena from the large corpuses of Croatian (hrWaC 2.2) and English (enTenTen13) using the SketchEngine API and UDPipe (http://ufal.mff.cuni.cz/udpipe) tokenizer and parser. The psychological domains are extracted using the syntactic methods of paradigmatic similarity score for the co-occurrences in the coordinated construction [x and y] for nominal lexemes Sketchengine platform (https://the.sketchengine.co.uk) that function as a detector of entities connected with logical operators. Using graph algorithms for community detection the lexemes in the coordinated linguistic constructions are classified for their syntactic- semantic domains. The third phase examines the ontological status of the lexemes and superimposes logical inferences on the semantic-syntactic constructional relations of the language specific knowledge. This empirical approach sets the dynamic systems theory as the epistemological basis for studying ontological questions of the syntactic- semantic relations expressed in language, its metaphoricity, dynamic network relationships, nonlinearity, emergence, complexity, hierarchy, ontological contingency and congruence of the conceptual organization of psychological concepts (Larsen-Freeman 2015).

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo, Filologija



POVEZANOST RADA


Projekt / tema
HRZZ-UIP-05-2017-9219

Ustanove
Filozofski fakultet, Rijeka,
Sveučilište u Rijeci - Odjel za matematiku