Telstar

IRM10 – from reference management to real-time impact metrics

Victor Henning is the last presentation of the day (we close with a panel session). Victor says research is inherently social. Mendeley is built on this concept. Mendeley both helped and hindered by lack of library background. In fact there is a strong music background to those involved in Mendeley.

The Last.fm model – you have a ‘scrobbler’ which monitors everything you listen to and uploads details to your last.fm account. This means you can build recommendations and links based on your listening habits. Mendeley makes research data social – mashing up research libraries, researchers, papers and disciplines (as opposed to music libraries, artists, genres etc.)

Mendeley offers a free desktop interface, which interacts with your web-based account – you can also login to your account on the website. Desktop interface extracts metadata from pdfs which are uploaded – and then uses that to get the correct metadata – e.g. if there is a DOI). You can read and annotate papers within the desktop application. Allows you to turn existing pdf collection into a structured database.

Mendeley includes ‘cite as you write’ functions – plugins for Word and Google Docs – you can drag and drop from Mendeley to a Google Doc. Also supports ‘shared collections’ – synchronises across desktop applications – including annotations etc. On Mendeley everything is private by default (contrasts with CiteULike). Mendeley is a tool for collaboration – and more functionality is coming around this. Mendeley can sync with both CiteULike and Zotero. Also support and bookmarklet and CoinS.

Mendeley allows you to see ‘real time impact metrics’ – most read papers, most used tags etc. Mendeley looking at recommendations not just on collaborative filtering, but also on analysis of content – extracting keywords etc.

What could it mean for Impact Factor? There are lots of criticisms levelled against citation-based metrics – skewed distribution, wrong incentives to researchers (target only high-impact journals, write with a view to citation), studies find only 20% of papers cited have actually been read. Mendeley can measure ‘usage’ of document by each user – annotations, how often opened etc.. It can also measure reading time and repeat readings per paper. Since user data recorded as well Mendeley can break down statistics by academic discipline, geographic region, academic status (students, researchers etc.)

Some data – e.g. ‘most read’ already on Mendeley website – and being picked up by researchers. Mendeley are not bibliometricians – so they are going to open up the data via an API so that libraries, publishers, bibliometricians can do analysis.

Coming in the future – better collaboration tools – Group/Lab management online, document-centric discussion feeds – all accessible via API. Full-test search in Mendeley and other databases, statistics queries and library systems integration also coming soon. Will be able to do queries like “what is the most read paper for this tag in this geographic region”.