Home // SEMAPRO 2023, The Seventeenth International Conference on Advances in Semantic Processing // View article

Conceptual Semantic Evaluation Metric Using Taxonomy

Authors:
İlknur Dönmez
Nur Bengisu Çam

Keywords: Similarity measures; word alignment; taxonomy; conceptual similarity; sentence similarity.

Abstract:
The conceptual method is an important technique for calculating semantic similarity. In this study, we propose a taxonomy-based formula for calculating the conceptual similarity of sentences. The coefficients in the formula calculate how similar the noun words that make up the sentence ("verbs" and "adjectives" are also included) are to their most similar conjugates in the other sentence by considering the distance of these two words from their common ancestor and the position of the common ancestor in the ontology tree. We test our proposed metric in the English Semantic Textual Similarity (STS) benchmark dataset for semantic similarity. Although the labels of the dataset were not generated specifically for conceptual similarity, we were able to achieve 77 % accuracy in determining similar sentences using our proposed formula (which uses only noun types).

Pages: 1 to 6

Copyright: Copyright (c) IARIA, 2023

Publication date: September 25, 2023

Published in: conference

ISSN: 2308-4510

ISBN: 978-1-68558-108-4

Location: Porto, Portugal

Dates: from September 25, 2023 to September 29, 2023