Ontology-Based Explanation of Discovered Associations in the Domain of Social Reality

One of possible uses of domain ontology as prior knowledge in KDD is the generation of explanations for discovered hypotheses. We developed an ontology covering a subset of concepts and relations relevant for ‘municipal social reality’, and manually mapped these entities on the structure of sociological data; the data originated from a comprehensive opinion poll over citizens of the capital city of Prague. The LISp-Miner KDD tool was then applied on data, and the most conspicuous associations were matched with the ontology. Some of the extracted ontology structures seem to offer useful insight into the background of empirical associations, as high-level templates that can be instantiated to concrete explanations.