Mining fuzzy functional dependencies from quantitative data

Present a data mining method for discovering fuzzy functional dependencies from quantitative data. A number of algorithms for mining functional dependencies from a crisp relational data model have been proposed recently. The concept of functional dependency is based on the assumption that data are either fully dependent or fully independent. However, this assumption is too restrictive for real-world applications. Our paper thus proposes a method to extract fuzzy functional dependencies from quantitative data. The dependencies discovered contain not only the conventional functional dependencies but also semantic dependencies between attributes. The discovered dependencies call be used in the reconstruction of the conceptual structure of relations and applied further to the integration of knowledge bases and databases as well as reverse engineering