Softly focusing on data

A computational approach providing a focus for unsupervised, reactive data mining is suggested. In data mining, achieving focus is an important issue. This is because there are too many attributes and values in a real database to consider them all. A soft focus is suggested, as both the data and the focus product may be imprecise. An approach is suggested for unsupervised searching controlled by progressive reduction of cognitive dissonance. Both crisp and non-crisp data are subject to discovery. Soft completing tools are needed because of the need to granulize data and to establish crisp boundaries in a non-crisp world. Issues involve: coherence measures, granularization, user-intelligible results, unsupervised recognition of interesting results, and concept-equivalent formation.