Academia.eduAcademia.edu

Sistem Temu Kembali Informasi dengan Metode Vector Space Model

Abstract

The information retrieval system (IRS) now provides document search results with abundant results of documents (high recall) and low accuracy (low precision). The objective of designing IRS with Vector Space Model (VSM) Method is to facilitate users to search Indonesian documents. IRS Software is designed to provide search results with the optimum number of documents (low recall) and accuracy (high precision) with VSM method that users may get fast and accurate results. VSM method provides a different credit for each document stored in a database which in turns to determine the document most similar to the query, where the documents with the highest credits are placed on the top of the search results. The evaluation of search results with IRS is conducted under recall and precision tests. This study fascinatingly creates a system which can preprocess (tokenizing, filtering, and stemming) within computation time of four minutes forty-one seconds. The system can search the documents and display them in the search results with an average computation time of 1.5 seconds, with an average recall of 0.19, and an average precision of 0.54. The system is equipped with a credit for each document, and the value is positioned on which a user may easily search the Indonesian text.