In this thesis we have attempted to formulate a model for an ‘ideal’ on-line finding aid for one such group of users, for which the ‘straightforward’ traditional IR model is not a perfect fit: historians. As frequent scholarly users of archives and libraries, they display a more organic approach to information-seeking. Their behavior resembles the ‘Berrypicking’ model as described by Bates (1989), and involves a wide range of techniques not covered by the traditional model, like serendipitous browsing, name-collecting and citation chaining.
In order to come to such a model we have not only analyzed the user, but also the usage—the historians' usage of primary resource material and bibliographic (meta)data, to be exact. For this we have examined the Evans corpus, a set of over 40,000 printed primary source documents concerning early American history, politics and culture. Evans is a premier example of the kind of bibliographic data historians come across and use in their daily research practice.¹
We have translated our findings regarding user and data into a new proposed model—and accompanying algorhitm—for historical interface design. This model—dubbed the ‘Semantically-enhanced Berry Basket’—combines the flexibility and usability of the SocialWeb (orWeb 2.0) with the power and precision of the Semantic Web. The ‘cognition-augmenting’ capabilities of Information Visualizalizations (Card et al., 1999), when combined with powerful ‘berry’-suggestion mechanisms (powered by ‘Semantic’ triples) form a promising basis for a user/historian-centered on-line electronic finding aid for primary resource material.