Chiang Mai Journal of Science

Documents are written using a natural language, so there are various interpretations because of different understanding of words. Therefore, implementing a mechanism to interpret statements to have only one clear point is very significant to readers. However, there are a veriety of formats of a sentence, and some sentences consist of serial verbs that affect to the meaning. Thus, knowledge discovering about serial verb constructions (SVCs) is significantly vital before understanding the entire sentence. This paper proposes mechanisms to unclose knowledge concerning to the consecutive serial verbs by focusing on determining relations between serial verbs among surrounding words. These relations and syntactic information are represented into the semantic binding patterns. Then, patterns of SVCs are generated from semantic binding patterns with semantic classes of surrounding words. Furthermore, the semantic interpretations of SVCs can be initially classified by partial semantic information of surrounding words.