engleski

Application of the graph algorithms for the semantic-syntactic construction analysis

From the systemic perspective of construction grammar (Langacker 2008), syntactic constructions express semantic functions of the conventionalized conceptualizations. The paper shows the use of graph algorithms for analyzing and enriching the lexical data obtained from the network patterns of semantic-syntactic constructions, also known as Word Sketches on the SketchEngine service (Kilgarriff et al. 2010), in a corpus. The study of enTenTen and hrWac corpora explains computational tools for API data harvesting procedures from SketchEngine using the Py2neo and Neo4j graph property database (Webber and Robinson 2018), the creation of the data model for storing lexical items and constructions that can be queried, analyzed using the graph algorithms for intersection, shortest path, centrality, community detection and visualization. The paper particularly demonstrates the network analysis of the syntactically connected lexemes within a textual corpus for identifying semantically related lexical items and semantic applications of the coordination (and/or) and identification (is_a) grammatical relations. The network analysis of these syntactic relations produces a valuable functional insight into the conventionalization of the lexical use, their semantic domains, polysemous structure of related domains, metonymy and metaphoricity. Sažetak Iz sistemske perspektive konstrukcijske gramatike, sintaktičke konstrukcije izražavaju semantičke funkcije konvencionaliziranih konceptualizacija (Langacker 2008). U radu je prikazana upotreba graf algoritama za analizu i obogaćivanje leksičkih podataka dobivenih iz korpusnih uzoraka sintaktičkih konstrukcija, također poznatih kao Word Sketches kroz uslugu SketchEngine (Kilgarriff i sur., 2010). Studija korpusa enTenTen i hrWac prikazuje računalne alate za procedure prikupljanja podataka sa SketchEngine API-ja koristeći Py2neo i Neo4j graf bazu podataka (Webber i Robinson 2018), stvaranje modela podataka za pohranu leksičkih predmeta i konstrukcija za stvaranje upita, graf algoritamsku graf analizu presjeka, najkraći put, centralnosti, detekciju i vizualizaciju leksičkih zajednica. U radu je oprimjerena graf analiza sintaktički srodnih leksema unutar tekstualnog korpusa za identificiranje semantički povezanih leksičkih predmeta i semantičkih aplikacija koordinacije (i / ili) i identifikacijskih (x_je_y) gramatičkih odnosa među imenicama. Mrežna analiza tih sintaktičkih odnosa daje vrijedan funkcionalni uvid u konvencionalizaciju leksičke uporabe, identifikaciju semantičkih domena, leksičku polisemnu strukturu, metonimiju i metaforičnost. References/Reference: Langacker, R., & Langacker, R. W. (2008). Cognitive grammar: A basic introduction. OUP USA. Kilgarriff, A., Kovář, V., Krek, S., Srdanović, I., & Tiberius, C. (2010, July). A quantitative evaluation of word sketches. In Proceedings of the XIV Euralex international Congress (pp. 372-379). Webber, J., & Robinson, I. (2018). A programmatic introduction to neo4j. Addison- Wesley Professional.

emotion, graph, construction grammar, emocnet

nije evidentirano