Access Data Mining: A new foundation for Added-value services in full text repositories

This paper describes the results of a breakout group at the 6th Open Archives Initiative Workshop (OAI 6) at Geneva in June 2009, also known as the CERN workshop on Innovations in Scholarly Communication. The breakout group focussed on the following issues: Users have many different needs and interests. Sometimes they are exploring the unknown at other times they would like to revisit some document vaguely remembered. Bibliographies, compilations of highly frequented works, lending records and many other methods were and will be employed to guide researchers towards the publications sought after. In the realm of electronic publications user behaviour can be observed in new ways. For example it is possible to track the browsing path of a visitor, a user's history is no longer confined to objects actually lended. Furthermore metadata describing and identifying the documents is obtainable just as easily. Many people are convinced that the combination of these types of data can yield great results, simplifying library searches, shedding light on the shadows of the deep web, or more generally speaking: Giving the user what he really needs. Two of the most outstanding applications of this paradigm are Amazon Recommendations and Google Search String Recommendations. Both are implemented to some extent in some repository solutions, but there is no doubt, that there are other services of which no one has thought before. The breakout was divided into four sections: 1. Free production (brain storming) of -preferably data based- possible Added-value Services 2. Integration of brainstorming results with ideas gathered in advance by the moderator 3. Estimation of the utility of the elements in the combined set of possibilities 4. Critical evaluation of the possibilitie