Actions Ontology System for Action Rules Discovery in Mammographic Mass Data

摘要

Actionable knowledge is a golden nugget within the data mining research field. Action rules describe possible transitions of objects in an information system -from one state to another more desirable state, with respect to a distinguished attribute. In this paper we propose an improved method for generating action rules by incorporating an additional ontology layer on top of the information system. It contains nodes of higher-level actions knowledge, which are linked with individual terms at the lower levels. The system shows the likely changes within classification attributes, with respect to a decision attribute of our interest. We experiment with Mammographic Mass DataSet in attempts to re-classify tumors from malignant to benign. In addition to medical domain, application areas include financial, and industrial domain.