Comparison of Chi-square Feature Selection and Singular Value Decomposition for Medicinal Plant Document Retrieval System

Abstract

This study analyzes the performance of the chi-square feature selection and Singular Value Decomposition (SVD) document retrieval system medicinal plants. The method is to find an effective document retrieval system based on queries to produce documents relevant to the user. Chi-square selection techniques make the selection based on the features that are important words that are used to create a vector space models. The model consists of several documents represented as vectors of the frequency of occurrence of features. SVD feature selection technique using key words found in each document. Important words become terms of the term-document matrix. Indonesian language document medicinal plants totaling 132 documents used in the chi-square retrieval and SVD. Retrieval and SVD chi-square test with 29 non-semantic and semantic query. The result is a comparison of the performance of the retrieval of both methods on non-semantic and semantic query.