Designing automated help using searcher system dialogues

This research utilizes a cognitive model of interactive information retrieval seeking to improve the performance of Web information retrieval (IR) systems. Building on the stratified model, we define interactions at the surface stratum that shed light on the cognitive, affective, and situational strata during the information retrieval process. We propose that one can utilize these interactions to improve the design information retrieval systems. This paper presents the development technique used to modify an existing IR system that monitors these interactions and, using associated assumptions about situational relevance, recommends search tactics to the user. The result is an increase in the performance of an IR system as measured by precision. The system design and results of an evaluation are presented. Research thus far indicates that user - system interactions at the surface stratum can be used to improve system performance. Using these interactions, one can develop the stratified model to a level of granularity useful for the design of Web IR systems.