Utilizing LOD Relationships and FOAF Vocabularies for Top-N Recommender System

• Published in: 2021 1st Babylon International Conference on Information Technology and Science (BICITS) pp. 98 - 103
• Main Authors: Mahdi, Ahmed M., Hadi, Asaad S.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 28.04.2021
• Subjects: Cloud computing; DBpedia; Feature extraction; FOAF; Linked Open Data; LinkedMDB; Motion pictures; RDF; Recommender Systems; Resource description framework; Semantic Web; Semantics; SPARQL; Vocabulary; Web sites
• DOI: 10.1109/BICITS51482.2021.9509914

The World Wide Web is transitioning from a Web of hyper-linked documents to a Web of linked data. A large amount of Resource Description Framework (RDF) data was published in publicly available datasets and connected to create the so-called Linked Open Data cloud. The semantics embedded in the Linked Open Data (LOD) can be utilized to enhance the recommender systems (RSs). Limited content analysis, cold-start and data sparsity are well-known issues in traditional RSs, which occurs when few or no features describe the items or no ratings to achieve a recommendation task. Because the LOD cloud contains many features, the knowledge encoded can help resolve this issue. The proposed system's main idea is to design a knowledge-based recommendation system that uses semantic features extracted from multiple datasets in LOD. Our approach will generate recommendations depending on direct and indirect relationships between resources. The results will be ranked according to their similarity score with the input before presenting them to the user. The results of our approach were comparable to the results of the Internet Movie Database (IMDb) website. Furthermore, an experimental evaluation using the MovieLens dataset was conducted. The results were encouraging and stimulating further research in this particular field of study. The usage of different kinds of relations in LOD can enhance the accuracy of the recommendations.