Boosting the Performance of Hybrid Semantic Ontology-based model using Density Rough k-Modes Clustering Technique on Library Book Recommendation System
Moh. Iqbal Zuhdi Husaeni, Noor Ifada, Fika Hastarita Rachman · 2023
This paper is a comprehensive study of implementing the Hybrid Semantic Ontology-based (HSO) model on the library book recommendation system. The study does not merely focus on implementing the model but also on boosting its performance. Initially, HSO consists of three modules: Collaborative Filtering (CF), Content-based (CB), and Hybrid (HB). Our proposed method adds another module, i.e., implements the Density Rough k-Modes (DRk-M) clustering technique prior to the CB Module in the HSO model. In this case, our method comprises four modules: CF, Clustering, CB, and HB. The CF module implements a matrix factorization algorithm to compute a book frequency circulation prediction score. The clustering module groups books based on their keywords. The CB module implements the ONTO Semantic Similarity (ONTO) algorithm to compute the semantic similarity of a book towards others based on its ontology in a cluster. Afterward, the HB module combines the scores calculated in the CF and CB modules as a hybrid book frequency circulation prediction score. A series of experiments using a real-world library dataset shows the very significant outperformance of our proposed method compared to the original HSO method. In terms of the Normalized Discounted Cumulative Gain (NDCG) evaluation metric, the average increase percentage of our proposed method towards the original HSO from Top-l to Top-20 lists of library book recommendations is 89.06%.