EN BG

Semantic Network with a Wide Range of Semantic Relations



Period: 2017-2019

Type of project: collective

Funding: National Science Fund, Grant Agreement ДН 10/3 from 14/12/2016



Principal Investigator: Prof. S. Koeva, Ph.D.

Participants: Prof. S. Koeva, Assist. Prof. S. Leseva, Assist. Prof. T. Dimitrova, Assist. Prof. M. Todorova, Assist. Prof. Valentina Stefanova, I. Stoyanova, B. Rizov, D. Hristov, M. Yalamov; prof. Tinko Tinchev (Sofia University), prof. Maciej Piasecki (Wrocław University of Science and Technology).

Abstract:

The project offers fundamental scientific research in the field of natural language semantics. Its main objective is to enrich the semantic networks with new semantic relations, which have not yet been the focus of either lexical semantics, nor known semantic networks. The Bulgarian WordNet (Koeva 2014), a lexical-semantic network for Bulgarian, will be used as a starting point for research.

Semantic networks are a form of knowledge representation that defines semantic relationships (relations). Semantic networks are a directed graph (conceptual graph) whose nodes are concepts or facts (names, dates, etc.), and the arcs between them represent different semantic relations. Examples of such networks are BabelNet, Wikipedia, DbPedia, Wikidata, WordNet, etc., and there is a trend to constantly enrich and integrate them together.

The main goal of the project is to enrich WordNet with a dense network of semantic relations. Enrichment will be achieved by presenting conceptual frames encoding semantic relations in a generalised way. A relatively small number of frames will correspond to a very large number of relations, as conceptual frames set patterns in terms of word classes.

The objectives of the research are:

a) To define a conceptual framework describing the relationships betweem: a semantic relation, a verb synset of a given semantic class, and the set of noun synsets belonging to appropriate semantic class(es). Our hypothesis is that there are regularities between semantic primitives and relations.

b) To construct conceptual frameworks that model the relationships between a semantic relation, noun or participle synset of a given semantic class, and the set of noun synsets belonging to appropriate semantic class(es). Our hypothesis is that the defined conceptual models for compatibility between a semantic relation, a verb synset of a given semantic class, and the set of noun synsets belonging to a given semantic class are transferable to the corresponding derivative participles and nouns (observing some systematic constraints).

Copyright © 2015 Department of computational linguistics. All rights reserved.