A New Approach for Arabic Named Entity Recognition

A Named Entity Recognition (NER) plays a noteworthy role in Natural Language Processing (NLP) research, since it makes available the detection of proper nouns in unstructured texts. NER makes easier searching, retrieving, and extracting information seeing as the significant information in texts is usually sited around proper names. This paper suggests an efficient approach that can identify Named Entities (NE) in Arabic texts without the need for morphological or syntactic analysis or gazetteers. The goal of our approach is to provide a general framework for Arabic NE recognition. Within this framework; the system learns the recognition of NE automatically and induces NE systematically, starting from sample NE instances as seeds. This method takes advantage from the web, the approach learns from a web corpus. The seeds are used to identify the contexts in the web denoting NE and then the contexts identify new NE. Thorough experimental evaluation of our approach, the performances measured by recall, precision and f-measure conducted to recognize NE are promising. We obtained an overall rate of F-measure equal to 83%.