Authors: Mohammed Nejja; Abdellah Yousfi
Addresses: TSE Team, ENSIAS Mohammed V University, Rabat, Morocco ' Eradiass Team, FSJES Mohammed V University, Rabat, Morocco
Abstract: This paper aims to shed light on a mechanism that will be used to exploit topical context information improving the accuracy of the automatic spell checking system. This study aims at solving the problem encountered by the auto correct spell checking system, which resides in the fact that the requested solution might be located in the last position. We have implemented a set of techniques in order to build a context oriented spelling corrector and by the end of this work the designed corrector will essentially be using a dictionary that contains a distribution of probability of a word occurrence in various contexts. This latter is constructed by bringing a collection of documents available via the internet.
Keywords: Arabic natural language processing; ANLP; automatic spell checker; context; N-gram; Levenshtein.
International Journal of Artificial Intelligence and Soft Computing, 2017 Vol.6 No.1, pp.56 - 74
Received: 26 Dec 2015
Accepted: 08 Nov 2016
Published online: 21 May 2017 *