Efficient Discovery of Differential Dependencies Through Association Rules Mining
Differential dependencies (DDs) extend functional dependencies (FDs) to capture the semantics of distance among data values. To mine DDs in a given relation is, thus, more challenging as the more general definition of DDs creates: a combinatorial large search space; and hugely sized minimal cover sets of DDs. This paper proposes a simple, yet effective and efficient approach to mine DDs in a given relation. We study and present a link between DDs and association rules (ARs): paving way for the adoption of existing ARs mining algorithms in the discovery of DDs. Furthermore, we propose a measure of interestingness for DDs to aid the discovery of essential DDs and avoid mining an extremely large set. Finally, we show the efficiency and scalability of our solution through experiments on three real-world benchmark data sets. The results indicate that our discovery approach is efficient and scalable.
KeywordsAssociation Rule Distance Relation Minimum Support Association Rule Mining Minimal Cover
Unable to display preview. Download preview PDF.
- 2.Borgelt, C., Kruse, R.: Induction of association rules: apriori implementation. In: 15th Conference on Computational Statistics, pp. 395–400 (2002)Google Scholar
- 5.Fan, W., Jia, X., Li, J., Ma, S.: Reasoning about Record Matching Rules. PVLDB 2(1), 407–418 (2009)Google Scholar
- 8.Koudas, N., Saha, A., Srivastava, D., Venkatasubramanian, S.: Metric functional dependencies. In: 25th International Conference on Data Engineering, pp. 1275–1278. IEEE Computer Society (2009)Google Scholar
- 14.Song, S., Chen, L.: Discovering matching dependencies. In: 18th ACM Conference on Information and Knowledge Management, pp. 1421–1424 (2009)Google Scholar