Project Title: A hybrid recommender system for multimedia content
20 word summary: Combine content and user-based approaches to improve recommendations of text, audio, and video content in BBC products.
Length of internship project: 3 months (June - September 2019)
Stipend: £1500/month (£4500 total)
Research/Activity to be undertaken
We want to understand how to select content from millions of articles, audio samples, and videos and surface it to our audiences for unique and personalised experiences.
With this project we aim to investigate how to develop a recommender system that combines content and user data to overcome the challenge of cold-start items, a prominent challenge in the domain of ‘news’ recommendations. The recommender system should learn from user interactions with various media types such that the end application could easily navigate between text, audio, and video items of similar content and context.
Machine learning methods will be used to extract features from different content types and information retrieval tools will be applied to evaluate the space of similarity. The final model will be compared with recommender systems currently in use by BBC products. This is part of a greater effort at BBC Datalab to develop machine learning systems that are focused on, and empowering, the user.
How does this project fit with BBC’s strategy
By the summer of 2019, Datalab will have developed a Machine Learning platform to facilitate data science work in the BBC. This internship is an opportunity to test and feedback on the platform and make improvements that could benefit other Data Scientists in BBC. What is more, as we are moving towards Universal Recommendations, research on hybrid recommenders that can serve multimedia content will become vital for many applications including News, Sounds, and iPlayer. This project is an opportunity to learn at a small scale before we take over recommendations for complex systems that serve millions of users.
This project fits into the Universal Recommendations initiative and aligns with the OKRs of Datalab. Improving our recommender systems means that our audiences can have better and personalised experiences when using our online products or mobile apps. This is in line with the goal to reinvent the BBC for a new generation and increase engagement with BBC content.
Python programming skills
Research experience on the topics of Machine Learning, Information Retrieval, Recommender Systems, or Multimedia
Knowledge of cloud computing services (GCP) and large data processing is a plus
Good communication and presentation skills
Knowledge sharing opportunity
Increase our knowledge on recommender systems including state of the art and benchmarking approaches
Extend our experience combining multimedia content in recommender systems
Test and improve our recently developed machine learning frameworks
Understand the challenges of hybrid recommenders before applying knowledge to large-scale projects
Expected Outputs and Key Deliverables
The expected outcome of this project is a new hybrid model that could be benchmarked against, and replace, existing recommender systems currently used in BBC products such as the BBC+ app.
Besides the model itself, a presentation and a blog post explaining how the model works are expected to be delivered by the intern upon completion of this project. The presentation will be used for internal communication and documenting whereas the blog post will be shared with external communities.
Depending on interest and time from both sides, the blog post could be extended into a research publication submitted to a conference or journal on the topic of recommender systems and multimedia (eg RecSys, ACM Multimedia, or ACM SIGIR conferences).
If you are interested in this BBC Internship opportunity, please submit a short (<500 word) statement explaining how your skills match the proposed research project to firstname.lastname@example.org by 11.59PM on Friday 29th March 2019.