Mobin: A knowledge-based morpho-syntactic parser for Arabic

摘要

The existing systems for parsing Arabic texts morpho-syntactically have heavily been based on databases of lexical entries common in Modern Standard Arabic generally used in the news. Despite some computational efficiency of the approach, the limited coverage of such systems has made them inapplicable for the registers other than the news, specifically Classical Arabic, the language of Arabic literature. Moreover, as the rapid change in any language would require, these databases are in need of continuous upgrading. And still more, the tagged corpora generated by these systems need massive morpho-syntactic disambiguation. To resolve these inefficiencies, Mobin has been developed to bootstrap morpho-syntactic tags in Arabic corpora in an innovative approach relying mainly on the vocalism in the Arabic texts. The implementation and evaluation of this knowledge-based parser for Arabic diacritized texts have proved it is highly effective and efficient.