Recommender Model Development

Popular content-driven websites like YouTube, Vimeo and Soundcloud receive a large amount of content annually and are visited by billions of users world-wide. However, the majority of the content on these sites has little to no structure. For example, many videos on YouTube are only found in the personal playlists and have virtually no user interaction or content data. Consequently, while many of these videos could be of high interest to the YouTubes community, a lack of reliable information makes it very difficult for recommender systems to surface them. This increasingly leads to the unwanted phenomenon where a tiny fraction of content goes viral, and the rest is never seen by anyone except the creator and his/her small network of friends. Milq aims to solve this problem through collaboration.
The intern is going to research and develop a recommender system by leveraging techniques from machine learning and information retrieval domains. The main goal of the project is to build and deploy a production-level system that would provide accurate personalized recommendations to every user on Milq, which is itself designed to be used by everyone seeking video or audio entertainment or to engage in a cultural experience.

Faculty Supervisor:

Richard Zemel

Student:

Partner:

Layer 6 AI

Discipline:

Computer science

Sector:

Information and Communications Technology; Technology; Entertainment and Media

University:

University of Toronto

Program: