The Mendeley Data Science team have been busy attending some important events around the world. One of them has been JCDL 2014, the most prominent conference in the Digital Libraries arena. The conference looks at many of the problems we’re tackling at the moment, such as article recommendations and the best ways of automatically extracting information from research articles.
Maya Hristakeva, Senior Data Scientist at Mendeley, was particularly excited about the various approaches to topic modelling that were discussed at the event. “Topics were used as features for a diverse range of tasks, such as prediction of an author’s future citation counts, making personalised recommendations, search, author disambiguation, and creating more relevant citation networks, all features that make a direct impact to the research workflow on Mendeley.”
“We saw some really thought-provoking output come out of the JCDL14 proceedings such as Characterizing Scholar Popularity : A Case Study in the Computer Science Research Community. In JCDL’14” explains Kris Jack, Chief Data Scientist at Mendeley. “Some of the interesting research questions raised included one by Gonçalves, G. D., Figueiredo, F., Almeida, J. M., & Gonçalves, M. A. (2014) which asked whether it is possible to represent the popularity of a researcher using the number of readers that they have.”
It was also nice to see evidence in some of the papers presented that Mendeley readership is highly correlated with various measures of academic impact, such as h-index and publication venue importance,” says Mendeley Senior Data Scientist Phil Gooch.
Overall, this was a really valuable opportunity to connect with researchers who are working on similar problems to Mendeley, such as metadata extraction, recommendations, and citation/author/venue disambiguation, so we’re thinking about the idea of perhaps running an open challenge to focus this research into concrete output that could be of use in features for our users. If you have any ideas around that, do get in touch on Twitter with @_krisjack@mayahhf and @Phil_Gooch
Note: At Mendeley, we believe in dogfooding (it’s not as disgusting at it sounds, merely techy slang for using your own product to validate the qualities of that product…) so Maya, Kris and Phil took notes using Mendeley Desktop 🙂
Mendeley is supporting the 3rd edition of the International Workshop on Mining Scientific Publications, which will take place on the 12th September 2014 in London. The event will bring together researchers and practitioners from across industry, government, digital libraries and academia to address the latest challenges in the field of mining data from scientific publications.
“We’ve had a record number of high-quality submissions this year, so were really spoiled for choice in putting together the agenda, which combines long papers, short papers, demonstrations and various presentations. We also worked with Elsevier to engage directly with the research community, which is really fantastic.”
As part of that ongoing outreach, Gemma Hersh, Policy Director at Elsevier, will be giving a brief presentation and answering questions from the participants regarding the company’s recently updated Text and Data Mining policy, and how it can best support the evolving needs of the research community.
Last week I had the pleasure of travelling to Hong Kong to give two workshop presentations at the ACM Recommender Systems conference. The art and science of recommender systems have come some way since the first time that “users who like X also like Y” appeared on an e-commerce site on the internet, and this year’s conference attracted several hundred delegates from both industry and academia. Despite its close association with customer satisfaction and the commercial bottom line, as a research topic Recommender Systems occupies a tiny and somewhat recherché niche within the computer science discipline of Machine Learning, which centres on the idea that if you present a computer program with enough examples of past events, it will be able to come up with a formula to make predictions about similar events in the future. For a recommender system these events record the interaction of a user with an item, for example Alice watched Shaun of the Dead, or Kris read Thinking Fast And Slow, and the program’s predictions consist of suggested new books that Alice or Kris might like, or of other movies similar to Shaun of the Dead, and so on. In our products these scenarios correspond to Mendeley Suggest, currently available only if you subscribe to a Pro, Plus or Max plan, and to the Related Research feature which we recently rolled out to all users in Mendeley Desktop.
One challenge for anyone trying to build a recommender system is that it’s hard to tell whether or not your predictions are going to be accurate, at least until you start making them and can see how often your users actually accept your suggestions. As there is a huge space of possible methods to choose from – far too many to test every possibility on unsuspecting users – ideally we’d like to be able to figure how well each prediction formula (technically each mathematical model) matches reality before we get to that stage. If and how that might be possible was a recurring theme of this year’s conference, and the subject of my first talk in Hong Kong.
Surprisingly for a field that has now seen several years of quite intense research interest and hundreds of peer-reviewed publications, most practitioners remain highly sceptical of the results reported even in their own research. This made it particularly interesting to hear conference presentations from large tech companies such as Google, Microsoft, LinkedIn, Ebay, not to mention Chinese counterparts such as Douban, TenCent and AliBaba, which were new names to me but who also operate at colossal scale. These organisations have both the scientific expertise to develop cutting edge methods and the opportunity to test the results on significant numbers of real users. You might be surprised to learn quite how much sophisticated research has gone into recommending which game to play next on your XBox.
At Mendeley we use a great deal of wonderful open source software, and so we’re very happy that the work we did in the Data Science team for my other presentation at the conference also gave us a chance to give something back to the developer community in the form of mrec, a library written in the very popular Python programming library and intended to make it easier to do reproducible research on recommender systems, even if you’ll still need to test your new algorithm on real people to convince most of us that it actually works.