<![CDATA[Book classification is an important activity in library management because it determines the grouping of collections based on subject areas and facilitates information retrieval. Manual classification requires considerable time and accuracy from librarians, particularly as the number of collections increases and book subjects become more diverse. This study aims to implement the Support Vector Machine (SVM) algorithm with Term Frequency–Inverse Document Frequency (TF-IDF) weighting to assist book classification based on the Dewey Decimal Classification (DDC) and to develop a web-based book classification system as a decision-support tool for librarians. The research data consist of book titles, synopses, and DDC labels obtained from the National Library of the Republic of Indonesia (Perpusnas) catalog and other library OPAC catalogs available on the internet. The research stages include text preprocessing, vocabulary construction, TF-IDF weighting, Cosine Similarity measurement as a data-scope filtering process, and classification using SVM. The Cosine Similarity testing on 20 out-of-category data and 85 in-scope data showed that a threshold of 22% achieved an accuracy of 96.19%. Furthermore, SVM model testing on 235 testing data across 46 DDC classes achieved an accuracy of 96.17%, precision of 97.09%, recall of 95.90%, and F1-Score of 96.49%. These results indicate that the combination of TF-IDF and SVM provides good classification performance in determining the DDC class of books. The developed system also provides librarian validation, as well as model management and evaluation, allowing the classification results to be used as recommendations before being established as the final classification.]]>
Copyrights © 2026
