Implementation of LSI Method on Information Retrieval for Text Document in Bahasa Indonesia

Pardede, Jasman and Barmawi, Mira Musrini (2016) Implementation of LSI Method on Information Retrieval for Text Document in Bahasa Indonesia. INTERNETWORKING INDONESIA JOURNAL, 8 (1). ISSN 1942-9703

Full text not available from this repository.
Official URL: http://internetworkingindonesia.org/

Abstract

Information retrieval system is a system that is used to obtain the information based on user’s requirement. In this study, Latent Semantic Indexing (LSI) method is implemented in the system to search and to collect documents based on overall meaning of documents instead of individual’s word. Typical of documents that needs to be retrieved in the system are text document in *doc, *.docx, or *.pdf formatted. In the text preprocessing phase, Nazief and Adriani Algorithm is used to eliminate the affix (prefix, suffix, etc.) of a word and then match them in database root word. To evaluate the quality of information retrieval performance, time response, values of recall and precision are measured. Implementation of multithreading from ‘read document’ to stemming process is required in order to improve time responses. The result shows by using multithreading, the greater number of term in document collection gives the more efficient in required time response. In term of the required time response, the document collection in docx format is the fastest, followed by doc and pdf format. For 80 documents and beyond, the system produces an error “OutOfMemoryError” at the matrix decomposition process. This means that the greater number of document in the collection, the greater memory is needed to execute retrieval process.

Item Type: Article
Subjects: T Technology > T Technology (General)
Divisions: Karya Tulis Ilmiah
Depositing User: Asep Kamaludin
Date Deposited: 11 May 2018 04:01
Last Modified: 11 May 2018 04:01
URI: http://eprints.itenas.ac.id/id/eprint/81

Actions (login required)

View Item View Item