Staff project based in the Department of Oncology

Autumn 2015 round

For more information, please contact innovations@it.ox.ac.uk

ABSTRACT

Setting up a successful research collaboration remains heavily dependent on persona l recommendation. Despite the growth of social networking and user profiling, existing too ls for connecting researchers are inefficient and do not necessarily enhance opportunities for research collaboration.

Our proposal seeks to develop a virtual recommender system, using machine learning and content­ based filtering techniques, that provides two-way matching for researchers who are loo king to collaborate on a project. There are numerous benefits to taking this approach:

• reduced reliance on the creation of multiple user profiles to promote one’s work , which can
• be difficult to keep updated/complete and are often left to collect virtual dust
• more time efficient and reliable than existing online method of search-contact-connect­ collaborate
• ensures only researchers actively seeking collaboration are connected – less time wasted on searching for interested parties; less hassle for researchers from unsolicited contacts
• less bias and more novel connections – researchers no longer constrained by who they know/who they have heard of
• matching will be based on motivation and research foci that complement one a not her, thus paving the way for more cross-disciplinary connections

We envisage the initial recommender system will be “fixed” through an internet portal built into the Apereo Open Academic Environment (OAE). However, there will be opportunities to expand the utility of this project to other platforms, for example via mobile applications, and incorporate additional functions. The underlying algorithm may also be applied to other situations w here a virtual person-to-person recommender system would be beneficial.