Analysis of Contruction Indonesian Sentence Graph Using Knowledge Graph

Abstract

The syntactic analysis of a language has an advantage as compared to the semantic analysis in terms of the time used to process the language, that is faster than semantic analysis. But, if the syntactic analysis is not followed by its semantic analysis it tends to cause ambiguity. For example is an Indonesian sentence kucing memakan tikus mati. In Indonesian, this sentence has an ambiguity whether the cat died after eating the rat or the rat has died eaten by the cat. Therefore, semantic analysis is needed to explain the meaning of this sentence. In such a case the method of knowledge graph can be used to explain the meaning of the sentence. This method is a new approach for natural language understanding. There are 9 binary relationships and 4 frame relationships as components of the knowledge graph. Using this method, a word is represented by a word graph, and a sentence is represented by a sentence graph. Analyzing the sentence using knowledge graph needs a chunking rule. This rule has already existed in the structure of English and Chinese, but it has not existed in Indonesian yet. The objective of this research is to constructs a chunking rule in Indonesian sentences and a sentence graph in Indonesian with the knowledge graph method. The benefits of this research is to provide the rules of constructing Indonesian sentence graph and to create an Indonesian sentence graph. The result of this research is Indonesian chunking rule with 8 indicators, they are comma and period sign, pronoun, adverb, preposition, jumps, logic words, interlude of breath, and conjunction. The other result is Indonesian sentences graph that describes semantics aspect of the sentence.