Classification of Book Collections Based on DDC 23 Using Text Mining Algorithm at UNIDA Gontor Library

Abstract

The collection of books in a library is a means of information that has become the main actor as a supporter of the existence of a library. UNIDA Gontor library uses the 23rd edition of the Dewey Decimal Classification (DDC 23) classification system, as a reference for the classification numbering system for each book collection. However, in the classification numbering there is no automatic system that helps librarians in assigning classification numbering to each collection. So it is necessary to select a suitable model system to be applied to the automatic classification system. The data used in this study is in the form of blurb data on each collection of Indonesian public books in the UNIDA Gontor Library. In this study, four methods of text mining algorithms were applied to be tested and compared. The algorithm used in testing this research is Multinomial Nb, Logistic Regression, Random Forest, and Support Vector Classifier. From the test results, the highest accuracy results are the Support Vector Classifier algorithm of 72%, while the Logistic Regression algorithm is 69%, Random Forest 69%, and Multinomial Nb 59%. Further research is recommended to apply the support vector classifier algorithm into the UNIDA Gontor library information system.