Perbandingan Efisiensi Model Ruang Vektor pada Sistem Temu Kembali Informasi

Abstract

Information retrieval system is a system to represent, store, organize, and process informations. Discovered documents were ranked by vector space model . Normalization of the vector space models similarity consist of cosine, Jaccard, and Dice. This research aims to compare efficiency of three vector space models based on recall and average precision (AVP), computation time, and algorithm complexcity. A thousand document were used in this research. The result showed that each coefficient of vector space model yield equal value for recall and AVP. The measure of similarity in cosine coefficient vector space model better than Jaccard coefficient and Dice coefficient, in terms of algorithms complexity and 3.1% faster than Jaccard coefficient and 9.4% than Dice coefficient, in terms of computation time.