Search Engine Dua Bahasa Berbasis Kamus Menggunakan Sphinx Search

Abstract

Cross Language Information Retrieval (CLIR) systems allow users to input the keywords in their own language and then the system will retrieve the relevant documents that written in the other language from database. This research implements a CLIR system that will retrieve relevant documents in Bahasa Indonesia and English by entering the query in Bahasa Indonesia or English. In order to translate the query, we use bilingual dictionaries. For indexing and retrieval process, we use BM25 and proximity algorithms from Sphinx. Performance of the system is determined using recall and precision with maximum interpolation. In the evaluation we conduct comparisons among Bahasa Indonesia search engine, English search engine, and Bilingual search engine. The test results show that the performance of monolingual search engine is better than bilingual search engine. The influence factors are the amount of translations, the combination of the wrong translation, and the amount of the document collections