A profiling-based movie recommendation approach using link prediction

Recommendation with better accuracy is one of the major concerns. The most of the existing works focused on the user-movie ratings and the movie features for offering the solution. But in context of today's OTT platform, the consumers' (users) attributes are supposed to be available and need to be considered as one of the decision variables within the recommendation process. We have attempted to propose a better recommendation scheme that considers all these three inputs (user attributes, movie features, user-movie rating) as decision variables. The contribution is to prepare a user (movie) profile that represents an affinity pattern of the specific user in context of movie rating. The said profiling approach helps to create groups of the homogeneous users (in terms of movie rating) that in turn assists in the process of more accurate recommendation. The proposed concept is implemented through rigorous experimentation on benchmark data sets for necessary validation. Moreover, we have compared the proposed approach with the notable existing approaches and significant improvement is noted.