Constructing An Automatic Lexicon for Arabic Language

In this paper, we have designed and implemented a system for building an Automatic Lexicon for the Arabic language. Our Arabic Lexicon contains word specific information. These pieces of information include; morphological information such as the root (stem) of the word, its pattern and its affixes, the part-of-speech tag of the word, which classifies it as a noun, verb or particle; lexical attributes such as gender, number, person, case, definiteness, aspect, and mood are also extracted and stored with the word in the lexicon. A lexicon its a collection of representations for words used by a natural language processor as a source of words specific information; this representation may contain information about the morphology, phonology, syntactic argument structure and semantics of the word. A good lexicon is badly needed for many Natural Language applications such as: parsing, text generation, noun phrase and verb phrase construction and so on. Many rules based on the grammar of the Arabic language were used in our system to identify the part-of-speech tag and the related lexical attributes of the word [13]. We have tested our system using a vowelized and non-vowelized Arabic text documents taken from the holly Qur'an and 242 Arabic abstracts chosen randomly from the proceedings of the Saudi Arabian National Computer Conference, and we achieved an accuracy of about 96%. We discuss the factors behind these errors and how this accuracy rate can be enhanced.