Birte Christensen-Dalsgaard at Ticer: Intelligent / Next generation / Dynamic catalogue

Birte starts her presentation with the vision that libraries can develop intelligent systems that are able to follow you, knows your different profiles and knows where you are. She is not shy of data mining to achieve this objective.

Federated versus Integrated search
In the definition of Dalsgaard Federated search is something that Metalib does. i.e. Searching different information silos simultaneously and merged the results on a single screen. Federated search was nice solution, but ranking is lously,
With integrated search all content is harvest and indexed within a single system and search by users with any kind of tool. With integrated search you are able to rank in theory much better. However, it will not come easy. You have to balance the relatively “thin” metadata catalogue records and fulltext information. Where will the catalogue record be of a journal like Nature, which is a very important term in the life sciences. It remided me of an article by Tamar Sadeh (2006) which uses different definition than use by Birte.

Federated search is typically associated with:
• Database approach
• Queries
• Based on Z39.50 protocol
• Structured
• “Exact” match

Integrated search is typically associated with:
• Search engine approach
• Natural language
• Large Volume
• Statistical approach

In Denmark they have carried out a data mining experiment with library lending data to develop a recommender system. To their own amazement their privacy policy police did not object, but wherever you are trying to data mine and model data on users, privacy problems might crop up.

Interesting point she argues that we need different search systems for different research questions. A common search is a known item lookup, which is completely different from an explorative search on a new subject. Perhaps we need different search engines for these questions, and not expect one system to handle those very different questions.

Realizing that we actually need different search engines, we need to develop the library system as a modular approach.

Towards the end she gets back to the paradigm of Robin Murray: Synthesize, Specialize, Mobilize.

References:
Sadeh, T. (2006). Google Scholar versus metasearch systems. High Energy Physics Libraries Webzine(12). http://library.cern.ch/HEPLW/12/papers/1/

Christensen-Dalsgaard, B. (2008) The Intelligent catalogue. http://www.tilburguniversity.nl/services/lis/ticer/08carte/publicat/christensendalsgaard.pdf