تفاصيل الرسالة العلمية

Key phrases are words or phrases that indicate the main topics in a document. Key phrases also give the prospective readers a way to quickly determine whether the document satisfies their information needs or not. Using accurate methods for keyphrase extraction is a great demand because of the huge number of documents available electronically. Little efforts are achieved for documents written in Arabic language, andthe previous keyphrase extraction system has low precision values. In our work, we intend to design a framework for automatic key phrase extraction from Arabic documents with higher precision values. This thesis presents a system for Arabic keyphrase extraction (called FAKEAD) based on statistical features and C-value approach. Our proposed system consists of a number of phases,including, document pre-processing, noun-phrase extraction (candidate extraction), candidate feature extraction and ranking. The proposed model performance is evaluated using recall, precision, and F-measure. Our proposed FAKEAD system is accurate in extracting Keyphrases from Arabic documents in comparison with KP-miner system. Moreover, the results of applying FAKEAD to extract keyphrases from full documents and their corresponding summarized documents gives better results when extracting from full document