期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2019
卷号：97
期号：2
页码：681-691
出版社：Journal of Theoretical and Applied
摘要：In alliance to the emerging internet technologies and services, the field of questions answering was one of the most trending topics. It is being used in multiple applications ranging from search engines to smart and complicated home assistance devices. In this paper, we are proposing an enhanced method and system for question answering that serve Arabic language questions. This system provides accurate paragraph level answers that extract its information out of documents dataset in different fields. The proposed system uses Support Vector Machine (SVM), Single Value Decomposition (SVD), and Latent Semantic Index (LSI) to classify the query in two phases. The method has been tested on a set of queries in different fields (classes) against a documents dataset of size 10,000 documents in 10 classes. The testing shows promising and accurate output for each of the test cases. Average classification accuracy reaches 98% using document classification metrics.
关键词：Support Vector Machine (SVM); Latent Semantic Index (LSI); Question Answering System (QAS); Arabic Language; SVD
