Open AccessPengaruh N-Gram terhadap Klasifikasi Buku menggunakan Ekstraksi dan Seleksi Fitur pada Multinomial Naïve Bayes
Author(s) -
Esti Mulyani,
Fachrul Pralienka Bani Muhamad,
Kurnia Adi Cahyanto
Publication year - 2021
Publication title -
jurnal media informatika budidarma/jurnal media informatika budidarma
Language(s) - English
Resource type - Journals
eISSN - 2614-5278
pISSN - 2548-8368
DOI - 10.30865/mib.v5i1.2672
Subject(s) - trigram , bigram , computer science , artificial intelligence , pattern recognition (psychology) , naive bayes classifier , feature (linguistics) , mahalanobis distance , natural language processing , linguistics , philosophy , support vector machine
Libraries have the main task in the processing of library materials by classifying books according to certain ways. Dewey Decimal Classification (DDC) is the method most commonly used in the world to determine book classification (labeling) in libraries. The advantages of this DDC method are universal and more systematic. However, this method is less efficient considering the large number of books that must be classified in a library, as well as labeling that must follow label updates on the DDC. An automatic classification system will be the perfect solution to this problem. Automatic classification can be done by applying the text mining method. In this study, searching for words in the book title was carried out with N-Gram (Unigram, Bigram, Trigram) as a feature generation. The features that have been raised are then selected for features. The process of book title classification is carried out using the Naïve Bayes Multinomial algorithm. This study examines the effect of Unigram, Bigram, Trigram on the classification of book titles using the feature extraction and selection feature on Multinomial Naïve Bayes algorithm. The test results show Unigram has the highest accuracy value of 74.4%.