A Proposed Architecture for Cold Start Recommender by Clustering Contextual Data and Social Network Data

Abstract

Recommender Systems (RS) help users in selecting the apt items based on their taste from a pool of items. These systems are able to do a proper recommendation with the aid of Machine Learning algorithms. The context of a user plays an important role in recommending relevant and important product/item to a user. Social media networks are useful knowledge sources to elicit more ratings from new users than state-of-art active Learning strategies. If we are designing an RS for users whose tastes differ according to the current context (e.g., feeling), we can collect contextual data and social media information so that we will be able to recommend the right product or item. We can do this recommendation by using cross-domain RS, Selective Context Acquisition, and Implicit Feedback. This paper provides insights based on the state-of-the-art contextual data and social media environments in providing the cold-start recommendations and also propose the architecture for recommending the items to solve the cold-start issue.