A dependency analysis of Thai sentences for a computerized parsing system

Abstract

This study aims at analyzing Thai sentences by using a dependency grammar. Fifty sentences are selected as corpus. The analysis yields two types of representation : syntactic dependency tree and conceptual network. Thai categories and Thai syntactic cases are proposed to convert a linear string of sentences into a dependency tree. The construction of syntactic relations on a dependency tree is governed by three sets of priority : bottom-up priority, immediacy priority and probability priority. The dependency tree is then converted into a conceptual network by the use of case mapping and conceptual case constraints. The analysis is implemented in terms of rules on CUPARSE, a computerized parsing system. Rules for both syntactic and conceptual dependency analysis are written and organized into links. Ten links are used for syntactic analysis and four links are used for conceptual analysis.