Integration of Association Rule Detection with Rule-Based Ontological Support for Product Recommendation

Abstract

Recommender systems can increase the sales by suggesting to users additional products, that were selected by their preferences. Availability of efficient mechanisms for generating the adequate products and for capturing the eligible information are the most important challenges of Web users. The Apriori algorithm is the best-known association rules mining algorithm, whose objective is to find all co-occurrence relationships between data items. In order to obtain a well defined information in an open standard format, we propose, an ontology-based recommender system, which describes the item features in terms of semantic concepts. In this paper, a methodology that combines the Apriori algorithm with a domain-specific ontology is proposed. The proposed model transfers the association rules on custom OWLIM rules, and by using OWLIM semantic repositories the detected associations will become public and can be interconnected with other information from the Internet. In this way, new facts can be deduced, therefore new relationships and rules between product items will be generated.