Combining Methods for Detecting and Correcting Semantic Hidden Errors in Arabic Texts

摘要

In this paper, we address the problem of semantic hidden errors in Arabic texts. These are spelling errors occurring in valid words and causing semantic irregularities. We first expose the different types of these errors. Then, we present and argue the adopted approach, which is based on the combination of several methods. Next, we describe the context of our work and show the multi-agent architecture of our system. Finally we present the testing framework used to evaluate the implemented system.