Indian Journal of Science and Technology
Year: 2017, Volume: 10, Issue: 29, Pages: 1-9
Mugdha Adivarekar* and Vina Lomte
Department of Computer Engineering, RMDSSOE, Pune – 411502, Maharashtra, India; [email protected], [email protected]
*Author For Correspondence
Mugdha Adivarekar Department of Computer Engineering, RMDSSOE, Pune – 411502, Maharashtra, India; [email protected]
Recommender Systems are playing very crucial and vital role in day today’s life. People are very active on e-commerce sites as they get whatever they want at home. Some financial recommender sites like Money Control are getting popular due to their variety of sectors. These systems actually work on basis of Collaborative filtering model and apply knowledge discovery techniques for live interaction with person. E-commerce sites also provide top-N recommendations to users when they log in to system based on their previous shopping or surfing or interests. Hence collaborative filtering is most used technique over the decade. Objectives: 1. To find meaningful subgroups, formulate the Multiclass Co-Clustering (MCoC) algorithm and propose an effective solution to it. 2. Applying traditional and easily adoptable CF algorithm to merge Top-N recommendation results. Methods/Analysis: Collaborative filtering methods have been applied to different data like Sensing and monitoring data, financial data, and Electronic commerce and web applications. Findings: Simplicity, efficiency and Classification accuracy are most important feature provided by CFA. It is more natural to provide recommendations based on correlated user groups but it’s not mandatory that one user must have interest in other things that are liked by people in group. Novelty /Improvement: This approach can be considered as an extension of traditional clustering CF models. Multiclass co-clustering technique proposes to generate top-N recommendations by maintaining user-item interaction matrix and making clusters by matrix factorization.
Keywords: Collaborative Filtering, Information Retrieval, Multiclass Co-clustering, Recommender System, User Profile
Subscribe now for latest articles and news.