The Opinionated Recommender

Recommender Systems (RSs) are devices that are used to filter data to combat information overload and provide time saving measures to the user. While RSs have traditionally been done using a content or collaborative based approach, recent times have seen a surge in alternative approaches to try and alleviate some of the traditional problems found there such as the filter bubble, matrix scarcity and cold start issues. Many of these new approaches attempt to lever new sources to provide more accurate recommendations and offset some of these issues. In this paper we will outline some of the current flaws and propose a hypothetical system that will exploit external sources to improve upon the state of the art.