Inconsistent data repairs in database integrations

In the operations of multiple database integrations, one of the most prominent data quality problems is the existence of inconsistencies. Due to independence of data management among different information sources, there exist several possibilities for repairs of integrated data. In order to get good repairs and preserve the correct data, this paper introduces a new algorithm for repairing inconsistent data among multiple tables based on functional dependencies, which keeps integrity of source databases without deletions of tuples. The algorithm determines the tuples that are potentially updated by analyzing related attribute statistic measures, and is of objective, accurate and effective properties. A repairing system structure is presented, and some comparisons with other related work are given at last.