Positional Relevance Model untuk Pseudo Relevance Feedback pada Sistem Temu Kembali Berbahasa Indonesia

Abstract

In document retrieval, it is difficult to user to choose a proper query in order to get relevant information. This problem can be solved by reformulating the query. This research tried to improve the relevance of retrieval result by implementing positional relevance model (PRM) for pseudo relevance feedback (PRF) based on the proximity aspect and the position of a word in the document. The evaluation was conducted by calculating recall and precision. The results show that an 4.38% increase in precision and an 11% increase in recall have been obtained.