Bernhard E. Bürdek1, Maximilian Eibl2, Jürgen Krause2,3 1College of Design, Offenbach, Dept. of Industrial Design 2 Social Sciences Information Center, Bonn 3 University of Koblenz
In a time of growing information recourses powerful information retrieval systems with easy to use user interfaces are needed in order to enable users to cope with the information overload they are confronted with for example by the Internet or on-line library systems. Finding the right information is becoming a more and more difficult task. We claim that this task can be eased by providing the user with an user interface based on a visualization and employing different retrieval models depending on the ongoing database search.
Information retrieval systems are commonly subdivided in two basic approaches concerning the query-result dependencies: exact match approach and partial match approach. The exact match approach is based on queries formulated using the Boolean operators AND, OR and NOT. The document space is strictly divided in relevant documents matching the query and non-relevant ones. The advantage of the exact match approach is that very distinctive queries can be formulated. The main problem concerning the user side is that the concept of Boolean logic is hard to learn and error-prone to apply. The partial match approach calculates different values of relevance for each document. Thus, the result set consists of a list ranked by the assumed values of relevance. Since the user does not have to cope with Boolean operators but just has to enter search terms without conjunction, handling partial match information retrieval systems is much easier than exact match ones. Nevertheless, the user has barely a clue