Implementasi Multithreading Untuk Meningkatkan Kinerja Information Retrieval Dengan Metode GVSM

GVSM is one of the models of IR systems. GVSM divided into two processes i.e. preprocessing process namely reading the text (*.pdf, *.doc,*.docx), tokenization, filtration, stemming, and parse the query and the process of calculating the relevance of the document that has been done preprocessing with user query to get the value of similarity. In general, to determine of the effectiveness of IR application is determined by two factors, namely recall and precision [7,8,10]. According to Jarvelin and Ingwersen, the effectiveness of IR application is not only determined by the recall and precision, but rather the ability of IR applications in helping users to complete the search more effective and efficient. The effectiveness of an IR application is determined by recall, precision, time requirements, and reporting documents are presented to the user [7]. To improve the performance of IR applications in time requirements is implemented multithreading in the preprocessing process, i.e. stage 1 to stage 4 GVSM method. The result showed that the method GVSM able to rediscover relevant documents with 100% the precision value, to get the recall value which is equal to the same documents collection either with or without multithreading, but can save processing time by over 50 %.