Query Expansion for Khmer Information Retrieval

This paper presents the proposed Query Expansion (QE) techniques based on Khmer specific characteristics to improve the retrieval performance of Khmer Information Retrieval (IR) system. Four types of Khmer specific characteristics: spelling variants, synonyms, derivative words and reduplicative words have been investigated in this research. In order to evaluate the effectiveness and the efficiency of the proposed QE techniques, a prototype of Khmer IR system has been implemented. The system is built on top of the popular open source information retrieval software library Lucene1. The Khmer word segmentation tool (Chea et al., 2007) is also implemented into the system to improve the accuracy of indexing as well as searching. Furthermore, the Google web search engine is also used in the evaluation process. The results show the proposed QE techniques improve the retrieval performance both of the proposed system and the Google web search engine. With the reduplicative word QE technique, an improvement of 17.93% of recall can be achieved to the proposed system.