An Indonesian question analyzer to enhance the performance of Indonesian-English CLQA

摘要

The most common approach in Cross Language Question Answering (CLQA) is to employ question translation. The easiest way to conduct question translation is by translating the source question into target language using existing machine translation or bilingual dictionary. However, the error resulted by the question translation may influence the accuracy of CLQA. This research proposes an adaptation in Indonesian question analyzer in order to extend the English monolingual QA (Question Answering) framework into Indonesian-English CLQA. The monolingual QA used in this research is Open Ephyra, an open domain QA framework that is modular, extensible and language independent. This research is conducted by custumizing the Open Ephyra into CLQA and implementing it for the Indonesian-English CLQA. A better performance is shown by employing a question analyzer on the source language side than using the question translation on the target language side. It should be noticed that the question analyzer is also suitable for monolingual QA in Indonesian QA.