Recommendation generation using typicality based collaborative filtering

The rapid growth of information availability on the Web related to movies, news, books, hotels, medicines, jobs etc. have increased the scope of information filtering techniques. Recommender System is software application that uses filtering techniques and algorithms to generate personalized preferences to support decision making of the users. Collaborative Filtering is one type of recommender system that finds neighbors of users on the basis of similar rated items by users or common users of items. It suffers from data sparsity and inaccuracy issues. In this paper, concept of typicality from cognitive psychology is used to find the neighbors of users on the basis of on their typicality degree in user groups. Typicality based Collaborative Filtering (TyCo) approach using K-means and Topic model based clustering is compared in terms of Mean Absolute Error (MAE).