Design and Development of Recommendation Systems Using Multi objective Item Evaluation and Multiple Prediction Strategies with Contextual User Similarities

Abstract

With the advent of the digital world the recommendation systems (RSs) emerge as the next generation technology for assisting the target users in finding the relevant in-formation of their interest. Primarily RSs aim to improve users experience through personalized information when users are crowded with a vast pool of information and many alternatives. They also potentially minimize the negative impact of excessive in-formation overhead while filtering the users anticipated information. The traditional paradigms in RSs are content-based Filtering (CBF) and collaborative filtering (CF). In CBF new items are recommended based on their similarity to items already present in the users profile. On the other hand CF approaches believe that nothing is known about the items a priori. They employ the ratings that are assigned by the user to find other similar users (or items) based on their rating patterns. This generic nature of CF systems is the potential reason for their immense success and broad applicability.