New trends for building arabic language resources /
اتجاهات جديدة لبناء موارد اللغة العربية
Mohamed Abdelrahman Zahran Mohamed ; Supervised Amir Atyia , Mohsen Rashwan
- Cairo : Mohamed Abdelrahman Zahran Mohamed , 2015
- 87 P. : plans ; 30cm
Thesis (M.Sc.) - Cairo University - Faculty of Engineering - Department of Computer Engineering
Language resources are important factor in any natural language processing application. However, the language resource support for Arabic is not mature because the existing Arabic language resources are either scattered, inconsistent or even incomplete. To solve this problem, first, we automatically bootstrap a rich Arabic language resource leveraging the existing resources. Next, we build the largest statistical Arabic language resource, and then introduce a new technique to map this statistical Arabic resource to the English counterpart outperforming standard techniques in this task. Finally, using the new statistical methods we present a novel hoto here technique to link conventional Arabic language resources to English using cross lingual lexical substitution outperforming the state of the art system in this problem
Arabic language resources integration Natural language processing Text similarity