Title: A linguistic approach to short sentences keywords identification for a question answering system

Authors: Denis Araujo; Sandro José Rigo; Bruna Koch Schmitt; Alencar Hentges

Addresses: Applied Computing Graduate Program, Universidade do Vale do Rio dos Sinos – UNISINOS, São Leopoldo, Brazil ' Applied Computing Graduate Program, Universidade do Vale do Rio dos Sinos – UNISINOS, São Leopoldo, Brazil ' Applied Computing Graduate Program, Universidade do Vale do Rio dos Sinos – UNISINOS, São Leopoldo, Brazil ' Applied Computing Graduate Program, Universidade do Vale do Rio dos Sinos – UNISINOS, São Leopoldo, Brazil

Abstract: One of the aims of question answering systems is to identify which words are more relevant to understand the users' needs. Known approaches involve the identification of the users' intentions through a set of previously built related sentences. Some limitations of these approaches are the lack of flexibility and limited selection options. In this paper, we present an approach based on computational linguistics to identify the keywords in short sentences for question answering systems. The main contribution of our approach is related to the new way we use the information generated by the natural language processing tools to identify the keywords of the sentences, by profoundly exploring the linguistic information to select the keywords of the questions. Besides, we emphasise the generalisation and the simplicity of our algorithm. The efficiency of our method was proved by the performance of 0.9776 in precision, recall value of 0.9962, resulting in an F1 score of 0.9868 reached in the validation experiment using QALD-7 as a gold standard.

Keywords: question and answer systems; natural language processing; NLP; information retrieval.

DOI: 10.1504/IJWET.2019.105593

International Journal of Web Engineering and Technology, 2019 Vol.14 No.4, pp.367 - 382

Published online: 05 Mar 2020 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article