Pencarian Teks Bahasa Indonesia pada Mesin Pencari Berbasis Soundex

Abstract

Soundex algorithm is one way to rectify the misspelling or mistyping in writing a query on search engines. In this study, we will conduct empirical validation to establish the applicability of Soundex algorithm to Indonesian document. The algorithm used is the Fusion Soundex algorithm because it is a combination of Russell Soundex, Celko Soundex, and Fuzzy Soundex. The document relevance value is measured by BM25 weighting algorithm, while the generated code relevance value is by DiceCoefficient algorithm. The results show that single word query outperformed double or more words one with the average precisions 0.390 and 0.145, respectively. This is because only single word was considered.