Abstract—Nowadays, recommendation algorithms allow users to fulfill their desires more easily. It may be used to propose things in various fields, including e-commerce, medical, education, and more. E-commerce industry is where most research is happening to help people find what they want. A recommendation system provides users helpful information regarding their possible activities and interests. The serendipity problem occurs when people become bored with identical recommendations to their profiles. Serendipity offers users limited and predictable content without a systematic approach to delivering new and surprising insights. The user only receives objects that are highly correlated with what he is interested in. We saw in our previous research that novelty and diversity represent ways to diversify recommendations that users did not know they needed. However, there are several criteria to study to have serendipitous suggestions. After studying and analyzing the concept of serendipity, this research aims to challenge several metrics often overlooked concerning accuracy. In this paper, we propose a novel methodology, capable of analyzing the user preferences and extracting their disapprovals that are incorporated into the recommendation process. This paper describes an Ideal Recommender System based on Five Qualities called IRS5Q, which gives a nice surprise, implying that a recommendation should be unexpected yet helpful to the user. Experiments show that the algorithm can propose many products that each user will enjoy. The results of IRS5Q were evaluated against the recommendation results of the content-based filtering approach. The outcomes showed the efficiency of IRS5Q with the MovieLens dataset and its capability to predict more accurately than the alternative approaches. We take an improved approach to assisting users to get out of their filtering bubble, monotony and redundancy in the recommendations made by achieving more than 83% in the diversity metric, 77% in the unexpectedness metric.
 
Keywords—recommender system, content-based filtering, serendipity, over-specialisation, novelty, diversity, unexpectedness, relevance, utility


Copyright © 2023 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.