You are cordially invited by the COE/ICS Dept., to attend a Graduate Seminar, on the above given title, by Mr. Sadam Alazani, MS Student, Information and Computer Science Department, on Tuesday, 02nd April, 2013, at 02:30 PM, in Building 22, Room 132.
Abstract: This paper is a course project on natural language processing and it is supervised by Prof. Sabri A. Mahmoud. It addresses the identification of the author of unauthorized documents. We investigate the authorship attribution of English texts using several feature types (viz. vocabulary richness, function words and n-grams features). In addition, feature selection techniques are applied to reduce the dimension of the feature vectors. Three classifiers (viz. Euclidian Distance, Artificial Neural Networks, and SMO-Support Vector Machines) are used with carrying out both binary-class and multi-class classification methods. Several experiments are conducted to evaluate the effectiveness of the selected features and classification techniques on the selected corpus. The experimental results show that our system can identify authors efficiently. This work is a baseline for our future work on authorship attribution of Arabic texts.
About the speaker: Mr. Sadam Hussein Al-Azani, is from Yemen. He is in fourth semester of Master degree and his major is Information and Computer Science (ICS). He is interested in artificial intelligence area, especially natural language processing, pattern recognition and expert systems (or so-called rule-based systems).