This track focuses on recommendations to users about scientific papers that they might be interested in, using a data set that comes from the Mendeley system (www.mendeley.com). The aim is to share recommendation approaches and discuss issues like the types of scientific recommendation services that social research platforms like Mendeley could implement or the types of data sets that could help advance research around scientific paper recommendation. Submissions are expected to use the already published Mendeley dataset which came out after the 1st DataTEL Challenge of the 2010 Workshop on Recommender Systems in Technology Enhanced Learning (RecSysTEL).
The ScienceRec Track asks participants to use and evaluate their approaches in an off-line manner, but is also interested in proposals for relevant services, navigational interfaces, visualisations of recommendations etc. Thus it welcomes submissions that will combine the data set with the Mendeley API. It has to be noted that in the Mendeley dataset, the document IDs do not match up to the document IDs in the API. For privacy reasons, Mendeley cannot currently reveal the document IDs in the dataset, as this will make quite obvious who’s libraries it is making public (creating privacy and data protection issues). Submissions will need to take into consideration this limitation, exploring (and showcasing) possible combinations of the data sources, demonstrating them using example, sample or simulated data, as well as outlining such kind of limitations and possible solutions – such as dataset licensing from the users themselves.