Title: An Arabic natural language interface for querying relational databases based on natural language processing and graph theory methods

Authors: Hanane Bais; Mustapha Machkour; Lahcen Koutti

Addresses: Team of Engineering of Information Systems, Information Systems and Vision Laboratory Faculty of Sciences, Ibn Zohr University Agadir, Morocco ' Team of Engineering of Information Systems, Information Systems and Vision Laboratory Faculty of Sciences, Ibn Zohr University Agadir, Morocco ' Team of Engineering of Information Systems, Information Systems and Vision Laboratory Faculty of Sciences, Ibn Zohr University Agadir, Morocco

Abstract: Nowadays, databases represent a great source of information. To extract information from these databases, the user needs to write queries using database query languages, such as structured query language (SQL). Generally, for using this language, this user must know the database structure. However, this task can be difficult for non-expert users. In that, the use of natural language to extract data from the database can be an important method. The problems in using natural language query are that it does not give any specification about the path access corresponding to the required data. In this paper, a model of a natural language interface for databases is presented. This interface allows the user to extract data from a database by using Arabic language and it obviates the need for users to know the database structure. Also, it can function independently of the database domain and it can to improve its knowledge base through experience.

Keywords: Arabic language; natural language processing; database; graph theory; Dijkstra algorithm; extended context free grammar; database structure; natural language interface; machine learning approach; knowledge base.

DOI: 10.1504/IJRIS.2018.092221

International Journal of Reasoning-based Intelligent Systems, 2018 Vol.10 No.2, pp.155 - 165

Received: 28 Apr 2017
Accepted: 06 Jun 2017

Published online: 11 Jun 2018 *

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