ASCM: Analysis of a Sequential and Collaborative Model for Recommendations

The recommender system can predict future lists of items based on the user’s sentiments and interactions. As the data is ubiquitous, we have a number of options available to make a selection. This is the scenario where the recommender system plays its role by narrowing the available options and making the selection process easy and effective. In this paper, we have proposed a sequential recommender system that considers the user interactions along with the items in sequential order and maps the sequences dynamically, in which the weight-age is given to current preferences and recent choices as explored by the user of the system. We have provided a detailed comparison of the proposed sequential recommender system with the commonly used matrix factorization model. The proposed recommender system is based on sequential dependencies and responds equally well to the cold start problem. We have obtained results using Movielens dataset, which reveal that the proposed recommender system is much faster with higher mean reciprocal rank (MRR) value.