Recommender System for Generic User Preferences for Online Content

Abstract

The population of users on the internet has increased steadily in the past years, and online content delivery platforms continuously generate ways in which content can be delivered to their users in a better and more efficient manner. Part of the necessity for this has to with the fact that the rate at which information is uploaded to the internet is overwhelmingly high, making it difficult in sorting through the numerous amounts of data available. For this reason, the rise of Recommendation Systems has been noticed. Much of the content that users decide to view is often delivered to them by recommender systems. These systems not only aid in helping users sort through data easily, but they also provide a great deal of revenue to most content providers. This research aims to identify a way in which current recommender systems can be lacking in certain functionality, and propose and implement a technology that can rectify this gap and be used for recommendations in ways which will be more beneficial to its users. The system developed through the research allows its users to have more direct control over their recommendations by allowing them to pick the data they want recommendations on, as well as allow files and data on their local system storage to be used to gather recommended content, something which cannot be done with commercial current systems. Testing performed on the resulting system has shown how this system can be beneficial to users.