Cognitive models in IR: Predicting task support for users' needs

Mark Ennis and Alistair Sutcliffe

The presentation outlines theoretical work on the cognitive activity underlying searchers' behaviour in information retrieval. This is to help address the mismatch between current information retrieval interfaces, and their associated functionality, and users' mental models of the search process. A cognitive model was developed based on existing theories of search behaviour and factors known to effect the retrieval process. The purpose of our work is to minimise the cognitive effort required to perform an information seeking task. This is to be achieved by linking the model of human action to what users need to know and the task support provided by an IR system (design requirements). In a step towards this model and in an attempt to validate IR design guidelines it was implemented in the COGENT environment. We discuss the use of computational modelling techniques as a method of increasing the understanding of complex tasks.

The later part of this talk concentrates on the implementation process and methods and functionality we would like to see provided for within the environment. The underlying question to be addressed is: does the support COGENT provide help validate complex cognitive theories.

Competence models ... Workshop 1 Specialised buffers