Building recommender systems for scholarly information

This paper details how the Mendeley Suggest recommender system has been designed and developed.

We live in the age of information overload, an no one feels this as much as academics and researchers. All of whom via to stay on top of the latest publications and finds. However, in a sector where the rate of publication have increase exponentially of the past few years means that it is even more challenging for research to find and stay on top of content within their fields. Recommender systems have been seen as a way of identifying relevant information to a user past on there historic tasks. This means that only content that is deemed relevant to a user in shown to them.

The paper presented at SWM (Scholarly Web Mining) details how the Mendeley recommender system has been developed, and evaluated. With the aim of recommending relevant new content to users based historic article that they have in their reading libraries. Additionally we detail how we deal with the issues of cold users, engaging them and on boarding them into Mendeley and Mendeley Suggest.


Building Recommender Systems for Scholarly Information from Daniel Kershaw

Mendeley Catalogue Page