data cleansing systems, data cleansing methods and data cleansing program

摘要

[Problem] in the data cleansing using the functional dependency of the Related Art, choose the modifications modification cost is minimized. In this case, when there are a plurality of modified tuples candidate correct cost is the same, it is not possible to choose the one most likely among them. A data cleansing system of the present invention includes a data-rule set storage unit 100, an exception cause the condition specifying unit 110, a tuple modifying section 120, and a amended data output unit 130. Estimating a cause conditions common multiple exceptions, by operative to determine the most probable correct tuple candidates by the estimated cause conditions, even impossible decision in modified cost only situation, the probable correct tuple candidate It makes it possible to elect. .BACKGROUND