Implementation of semantic analyzer in Indonesian text-understanding evaluation system

This paper proposes a strategy of semantic processing implemented in an Indonesian text understanding evaluation system. It uses component that already developed in Institut Teknologi Bandung consists of POS Tagger and Syntactic Parser. This research used syntax-driven semantic analysis technique by attaching semantic rules into associated syntactic rule. Semantic rule is used as base of λ-reduction operation, used to produce sentence's knowledge representation in first order logic's form. There is also semantic lexical attachment process, used to give a meaning of each words or phrases in sentence. Semantic processor is implemented in Indonesian text-understanding evaluation system. It has ability to evaluate a reader's understanding of a text by comparing the reader's understanding with computer's understanding. This system consists of text understanding module, question generation module, and understanding evaluation module. Text-understanding module is developed using semantic processor with addition of reference resolution to relate each sentences in text.