A computational semantic information retrieval model for Vietnamese texts. (4th June 2021)

• Record Type: Journal Article
• Authors: Do, Tuyen Thi-Thanh; Nguyen, Dang Tuan
• Is Part Of: International journal of computational science and engineering. Volume 24:Number 3(2021)
• Page Start: 301
• Page End: 311
• Publication Date: 2021-06-04
• Subjects: semantic information retrieval -- lexicon ontology -- semantic distance -- dependency graph; Computer science -- Mathematics -- Periodicals; Computer simulation -- Mathematical aspects -- Periodicals; Computational intelligence -- Periodicals; 004.015105
• Languages: English
• ISSNs: 1742-7185

Abstract

Semantic information retrieval system for text documents aims at retrieving text documents containing the similar semantic representation to the query. Semantic representation of text can be vector or dependency graph depending on the approach of semantic analysis. This paper proposes a model of semantic information retrieval for Vietnamese to retrieve similar texts to a query. In the proposed system, the semantic analysis is to identify the semantic dependency graph of sentences and the retrieving process computes the relevance of text document with these semantic dependency graphs. For identifying the semantic dependency graph of a sentence, the transformation rules are studied to apply on dependency parse using lexicon ontology for Vietnamese. For ranking retrieval results, the Jaccard-Tanimoto distance is applied to the ranking function. The evaluation shows that the proposed model has higher MAP (0.4045) than MAP of BM25 model (0.3825) and of TF.IDF model (0.3688).