WAY: An Architecture for User Adapted Access to Z39.50 Servers based on Intelligent Agents

Abstract

The work presented in this poster is based on the combination of three well-known paradigms of Computer Science applied to Digital Libraries: User Adaptive Interfaces, Information Retrieval and Intelligent Agents. Its objective is to define a model providing intelligent information access by means of an adaptive user interface. The model, called WAY (see figure 1), is supported by a web-based architecture and counts on the help of intelligent agents in charge of studying the user characteristics in terms of previous behaviours, in order to provide an adapted interface and to guide the user’s searching process through the servers available in the net. This model is currently being implemented in java using RMI (Remote Method Invocation) technology.

Concerning user adaptive interfaces, what our model proposes is to offer a different initial interface for each type of user. These initial interfaces will be defined according to the users classification, taking into account where the user is accessing the system from. Once the user begins to interact with a specific interface, it will go on changing depending on the actions performed by the user in two ways: the preferences shown explicitly by the user -selected options in the configuration of the interface- and the preferences shown implicitly -by systematic actions. The following examples illustrate these ideas:

• The type of user: in a university library environment, the possible types could be students, teachers and library staff, all of them with quite different goals and needs.

• The user mother tongue: if the user is connecting from Germany, he/she is likely to prefer a German interface.

• The user preferences: there are people who still prefer textual environments although their computers allow a good performance of visual ones.

• The user repeated actions: in a university, students usually are more interested in books than in journals, unlike teachers, whose main research tools are papers.