A LARGE-SCALE COMPUTATIONAL PROCESSOR OF THE ARABIC MORPHOLOGY, AND APPLICATIONS

Abstract

For a natural language, morphology is the very basic layer over which the higher syntactic and semantic layers are built. So, most of the software systems concerned with the natural text-based human-machine interaction are either direct or indirect candidates of reliable and efficient computational morphological processors. Through out all the significant contemporary natural languages, Arabic has the most elaborate though systematic morphological system. It is hence an excellent candidate of modeling without which any serious computational processing of the Arabic language, with all of its layers, is infeasible.

An industry-quality computational processor of the Arabic morphology­ called Morpho3 - along with a host of dependent applications as well as complementary utilities have been implemented. This thesis presents the theoretical basis upon which this morphological processor has been built as well as its domain of applications. Special attention was given to explaining how Morpho3 approaches the main problems that challenge any serious system for processing the Arabic morphology such as coverage, tolerability and ambiguity.

Morpho3 may be regarded as a demonstration of how a rule-oriented knowledge base and a statistical knowledge base can be married towards solving problems in computational linguistics.