The Haiku system provides a visual representation of discovery performance and interactions, a unique domain independent visual clustering algorithm, and a visual representation of general patterns and relationships within the dataset.
This work has shown Haiku as a useful and flexible tool for the visualisation of classification rules. However, Haiku is applicable to other forms of discovery, and is not limited to a single type or form of data.
The future use of prototypes could be used to increase the volume of data handled by the system .
Haiku provides a valuable step toward an interactive, symbiotic approach to to the integration of human and machine discovery skills.