Tuesday 8 April - morning sessions

The morning sessions represented UKSG at its best, concentrating on the theme of usage but with different aspects covered by all speakers.

The MESUR project: in search of usage-based impact metrics. (Herbert van der Sompfel, Los Alamos National Laboratory, Research Library)
MESUR is the first scientific approach to usage data & first scientific exploration of a new paradigm for scholarly assessment providing the infrastructure for future research about usage. MESUR is investigating lots of metrics. In order to construct usage metrics scenarios replicating how a user would use articles were set up and researchers constructed a network map of this usage. In May the project will release a java applet - the MESUR explorer prototype to visualise this map. Interpretations of the usage map reveals that users read widely, and that usage is linked to citations, but when they cite the stick to their disciplines. A citation-based map offers a less rich network with less connections and less disciplines.
While there are a lot of challenges to measuring usage, it is clear with the vast quantities of data available that this is a complex area but the final statement made clear the enthusiasm for the subject:

"Usage data totally rocks!"

The information-seeking behaviour of the virtual scholar: from use to users (David Nicholas, University College London)
Expanding upon the theme of user behaviour not usage, this speaker described The Virtual scholar project which addresses the concern regarding the disconnect when you measure usage but not user behaviour - “the user is king”.

With more information than ever we are able to investigate digital information footprints - which pages are viewed, for how long, what was viewed, when and repeated visits. This research has noted that there is a high demand for scholarly data/products and while Open Access articles may not be heavily used in the UK they are being heavily accessed by others, especially Asia, reflecting the good information brand UK research has. It is however more difficult to determine brands for authors, societies or publishers. The research has yielded interesting results: students constitute the biggest users but half the visits to publishers sites are made by robots (estimated to be about 50%) who mimic human information-seeking behaviour to gain entry.

To mimic human behaviour robots should a) be promiscuous- 40% of people don’t go back to the page b) bounce - view 1.3 pages from thousands and dip in and out of pages; difference between ages, students bounce more, scholars bounce less. c) flick ie channel hopping d) view- don’t look at article more than 2 minutes; if it’s a long article then we read the abstract & put away until it is not read - something I’m sure many could identify with! e) power-browse through titles at a huge rate of knots -spending more time reading short articles than long ones f) navigate - spend half the time navigating to content (The journey is much more fun than getting there). There are age differences and gender differences between searchers; older users tend to come back to the document, women are likely to return to read articles they have stored and even national differences.

The speaker used the adage that we are all librarians now, in fact we are eshoppers. An article from The Guardian was cited as reflecting this new information seeking behaviour - disciplined reading and studying is imploding - which raises the question, what do we think is good or bad information seeking? We need to show that good information seeking behaviour leads to good outcome. Access provision is not enough and he urges libraries to define best practice in terms of information-seeking behaviour.

Use & abuse of usage measures (Ian Bannerman, Taylor & Francis)
This session explored the reliability of usage date and referred to usage of Taylor & Francis journals to illustrate points. We assume usage data is consistent, credible & compatible & that usage would be a meaningful indicator of something. COUNTER was launched in 2002 to make usage credible. The release 3 of COUNTER is getting to grips with robots and data mining but weren’t felt to go far enough. Most known robots won’t get past access control to subscribed content but the unknown robots (such as crawlers within universities) distort the numbers and COUNTER 3 is not able to exclude these. However a point later from the floor was made that academics build robots in response to not being able to download information responsibly from publisher sites. Publishers need to improve accessibility and readability of content.

To test whether usage is being monitored a journalist found it possible to download thousands of articles and no alarm bells were raised by the publisher. The speaker also used the example of one article downloaded 6372 times (once every 9 seconds) by a Russian institution. COUNTER would regard this behaviour as a double-click & seen this usage as one hit but if the usage hadn’t triggered the double-click rule…There is a danger that measuring literature changes the literature ie the observer effect. Impact factors are also open to abuse, eg If the impact factor is published in December then authors can increase cites by publishing earlier in the year, or by encouraging fellow colleagues to download their article.

Can usage data can address the question of comparing like with like? There is a need to improve guidelines on detecting and blocking robots, carry out further research into factors influencing article downloads and librarians were urged to be cautious when interpreting usage data.