Sharing and innovating

Why did you start Mendeley?

The basic idea came about when my co-founder Jan Reichelt and I were doing graduate studies. Even though we were working in different fields we realised we were dealing with the same issues of information overload. We thought there really should be an easier way to manage our references – something like iTunes for PDFs where you could drag and drop the papers and extract the bibliographic information.

Most PDFs don’t make use of metadata, especially the older ones, so we had to develop algorithms to extract the information.

Our original idea was to license the software to people. Then we realised that if lots of people were doing this using the same system it could turn into a great collaboration tool if it’s networked and if we allow people to collaborate.

That idea drove us to take it more seriously and to get a third co-founder with a software background, Paul Föckler. We also got a fourth co-founder, Stefan Glänzer. He was the founder of Last.fm, a music service that uses audio metadata of preferences to create personalised radio stations. Through this process the company also aggregates people’s musical interests to create an enormous database of music preferences. This is similar to what we are doing. Glänzer used to be a guest lecturer of ours at business school so he knew us. He had also just sold Last.fm and was looking for a new challenge and introduced us to the Skype founders as investors.

We started the company in January 2008 in London and one of our first landlords was Michael Palin of Monty Python.

How has it gone so far?

The first public version was launched in December 2008 and we already have almost 300,000 people registered on the site or using the software. The five institutions where we have the most users of the service are: Cambridge University, Stanford University, MIT, Harvard and Imperial College.

Between these users they’ve uploaded 17 million papers into users’ private accounts and the number of papers we have is currently doubling every 12 weeks. People can use their private accounts to synchronise their libraries across their devices. Cite U Like users can synchronise their libraries with Mendeley too. We cover pretty much every area of research. Around 30 per cent of our users are biomedical scientists and the second biggest group is computer scientists.

We anonymously aggregate the bibliographic data from these papers and get usage statistics. We track whether people actually read and use the papers and how often they are shared. You can see the most widely-read papers and authors in each discipline and other trends.

What is the reaction from publishers?

The interesting thing is that before publishers talk to us they are concerned about our potential to be a sort of Napster for research, but it’s not really like that. Users have to invite people into their shared groups as contacts before they can share people. In a way it’s very like email or any internal online sharing system.

When publishers talk to us they are very interested in our recommendation algorithms and its distribution potential. We could provide users with recommendations of materials they don’t already have. We are setting up pilots with Oxford University Press and the Royal Society to do this.

Many publishers are not really happy with how impact factors are calculated or with the time lag. Even PLoS, which we see has a lot of readers, is not really represented well by impact factors. PLoS is one of the publishers interested in displaying our data.

What are you planning next?

We launched Mendeley as a free service and our philosophy is that once something is free we don’t charge for it, but we will start offering individual premium accounts to users. Bigger file capacity and group sizes above the current limit of 10, could be part of a premium account.

We get requests from the pharmaceutical industry for an enterprise version too. At the moment synchronisation between devices and group members happens on our servers, but these companies want all the resources they are using on projects to be shared inside their firewalls.

There is also interest in the usage statistics we are generating. Universities are interested in comparing their research output with that of other universities and getting metrics about research and authors.

We are looking at integrating Mendeley with institutional repositories such as e-prints and DSpace too. We are also collaborating with several European research universities and have just joined ORCID, an initiative set up by publishers for author disambiguation.

Opening our API will give other services the ability to tap into our database and build applications on top of our wealth of social information. We are talking with ChemSpider about mining for chemical structures and to geo-physicists in Stanford University, who want to mesh up our data with geographical data when they are in the field.

Interview by Siân Harris