1. UKM Learning and Research Repository
2. Master Thesis / Tesis Master
3. Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Title:	Shallow parser to recognize Arabic sentence types based on appropriate conjunctions
Authors:	Hanan Ali Abo-Abdullah (P50162 )
Supervisor:	Mohd. Juzaidin Ab Aziz, Prof. Dr.
Keywords:	Shallow Parsing NLP systems
Issue Date:	22-May-2014
Description:	Shallow Parsing is one of the important NLP systems to achieve many advanced NLP applications, such as Information Retrieval (IR), Information Extraction (IE), Question Answering (QA), and Automatic Document Summarization (ADS). This study concentrates on Arabic shallow parsing which will be used to recognize the types of Arabic sentences based on appropriate conjunctions that play an important role in the sentence. As a consequence, the main problem that will be addressed in this thesis is to identify the positioning of the conjunctions among the sub-sentences (parts) in Arabic sentences accurately. The first objective is to identify the conjunctions in an Arabic sentence and constitute the Arabic grammar rules of the parsing to identify those conjunctions. Then, we designed and implemented the Arabic shallow parser system to recognize the types of Arabic sentences based on the conjunctions that have been identified. This research will propose the production of a Rule-Based method, in order to achieve a shallow parsing task for Arabic sentences and the recognition of its types, as a Rule-Based approach has been successfully used in developing many other natural language processing applications. Our system has included two phases which are, shallow parsing and the recognition of Arabic sentence types (i.e., Simple, Complex and Multiple-Complex). In the first phase, the main Arabic phrases (i.e., NPs, VPs, PPs and CPs) are determined within two levels. In the second phase, a novel method of recognizing the types of Arabic sentences will be proposed in a single level. We have evaluated a Rule-Based approach that handles shallow parsing to recognize various types of Arabic sentences. The system was manually tested on 75 Arabic sentences; these cover many types of Arabic sentences (i.e. Simple, Complex, and Multi-Complex). The results obtained have achieved a 91.35% precision for the shallow parsing phase and an 89.33% precision for the recognition Arabic sentence types phase, which proves that the system provides good results and reflects the viability of this approach for recognizing the different types of Arabic sentences.,Master/Sarjana
Pages:	149
Publisher:	UKM, Bangi
Appears in Collections:	Faculty of Information Science and Technology / Fakulti Teknologi dan Sains Maklumat

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.