Enhanced semantic arabic Question Answering system based on Khoja stemmer and AWN

Abstract

Due to the great amount of information available on the web, Question/Answering systems have become a focus for researchers and users as well. This paper introduces a proposed design for an Arabic Question Answering system based on Query Expansion ontology and an Arabic Stemmer. A set of factoid CLEF and TREC questions used to evaluate the system. Improved results obtained using AWN as a semantic Query Expansion and Khoja stemmer as a stemming system. Two experiments conducted using AWN the first using one level of expansion and the second using two level of expansion. Three measures are performed: Accuracy, Mean Reciprocal Rank, and Answered Questions, and the obtained results are 35.5%, 20.2%, and 65.33% respectively when using one level of expansion. But when using two level of expansion we get 38.77%, 16.2%, and 65.55% respectively.