International Journal of Modern Engineering Research (IJMER)
Functional Dependencies (FD's) are recently extended as Conditional Functional Dependencies (CFD's) for cleaning relational data by supporting patterns of semantically related constants. Finding worth CFD's is an expensive process which needs more manual work. Using relations the authors are able to effectively identify data cleaning rules. Even the mining patter provided with new techniques, they in this paper provided three techniques for CFD discovery. First is a technique for closed mining item sets which is used for discovering constant CFD's ie's CFD's only with patterns which is essential to data cleaning and integration, which is referred to as CFDMiner. The other two algorithms are CTANe and TANE which are developed for discovering general CFD's.