iCFDMiner

Conditional functional dependency (CFD) has been verified to be more effective for checking data consistency than traditional FD, and there are quite a few algorithms of mining CFDs from a static database. However, records in a database are frequently added, deleted or modified in reality. Thus, relevant incremental algorithms are preferred in a dynamic updating database. To our knowledge, the study of incremental algorithms for mining CFDs are rare. In this paper, an incremental algorithm, iCFDMiner is proposed based on the batch algorithm CFDMiner, which is very popular for discovering constant CFDs in static databases. It is proved that iCFDMiner scales well with the size of the database, and all operations (adding, deleting and modifying). Experiments show that iCFDMiner outperforms CFDMiner in terms of running time and computing spaces in most cases.