Refiner++: A Knowledge Acquisition and Refinement Tool

Aiken, Andy and Sleeman, Derek (2003) Refiner++: A Knowledge Acquisition and Refinement Tool. Capturing knowledge from domain experts: Progress & Prospects, Sanibel Island, Florida.

Full text available as:

PDF - Requires Adobe Acrobat Reader or other PDF viewer.

Refiner+ is an algorithm than detects inconsistencies in a set of examples (cases) and suggests ways in which these inconsistencies might be removed. The domain expert is required to specify which category each case belongs to. Refiner+ infers a description for each of the categories and reports any inconsistencies that exist in the dataset. An inconsistency is when a case matches a category other than the one in which the expert has classified it. Refiner++ is a new Java implementation of the (LISP-based) Refiner+ algorithm. The original algorithm was modified in several respects. Refiner++ was designed to be an application that would allow a domain expert to enter data, manipulate it and examine it without the need for a knowledge engineer to act as an intermediary. At the time of writing, the Refiner++ system has been presented to three experts to use on problems in their domains: anaesthetics, educational psychology, and intensive care. The work is ongoing, but at this stage we have identified types of tasks Refiner++ is particularly good at solving, and some for which it is less useful. The application has met with great interest from domain experts, but as yet none seem willing to use it directly. Enhancements to the UI and the materials used to introduce a new domain expert to the conceptualisation task inherent in a Refiner++ session are under way. Shortly we plan to run a series of new studies with further domain experts.

Contact the site administrator at: hg@ecs.soton.ac.uk