Information-Based Language Analysis For Thai

Thai language is recognized as an isolative language having neither lexicon inflection due to word agreement and tense (as in English) nor obvious syntactic case marker (as in Japanese). The position of word in a sentence is the important superficial syntactic information for recognizing the meaning and the syntactic role. In this paper, we are going to describe the methodology and algorithm to cope with the mentioned Thai specific language phenomena. The rather fixed relation of word position and its syntactic role gives a well-formed pattern of a phrase. Therefore, localization of pattern analysis helps much in phrasal recognition and works well in lexicon disambiguation. In a sentence, the relatively less ambiguous in concept of words (variety of concepts) are consecutively determined to make up a bunch of concept. Then, according to the information retrieved from dictionary, subcategorization employed by the verb of the sentence will finally create the relation between those groups of concepts to build up a dependency structure to represent the meaning of the sentence. Besides the lexicon information from the dictionary, the grammatical rules are employed to identify the appropriate semantic relation between concepts with lexicon functional reasoning in the pair of provides and requires attribute.