Ekspansi Kueri Menggunakan Metode Semantic Similarity Retrieval Model (SSRM)

Abstract

The main objective of this study was to applied the method of semantic similarity in the proccesed of query expansion in information retrieval systems in the Indonesian language. Selection of partners in selecting candidates expansion will be used phrase pairs that have been made In research Kartina (2010) in which the phrase pairs with the largest value of similarity between the words that will be selected. Used by 10 groups of queries not clear who obtained manually and 30 groups of queries in 2000 agricultural document, document search results will be compared with the addition one-term of expansion, two-term of expansion, and three-term of expansion. The results of this study suggest five things. The first that the expansion of the query by added one term produces better results than the addition of two terms and three terms. The second that query expansion used by 10 groups of queries produces a higher precision values than using 30 groups of queries. The third that the used of the semantic similarity method produces better performance than the use of a thesaurus on the method of similarity retrieval Vektor Space Model (VSM). The fourth that the used of semantic similarity methods have not been able to produce better performance than the used of the method of conditional probabilities in the selection of expansion terms resulting from the translation of bilingual dictionary. The fifth that the used of the phrase paired in determining the candidate expansion terms can not maximize the search results using the method of semantic similarity of documents, because each pair selected does not necessarily have a semantic relationship with a given query.