Abstract
A system for cleaning a database (201) in which the database is partitioned into fragments (203) and each fragment is checked for violations of a data quality specification (205), then at least one data cleaning asset (such as a machine based data cleaning asset or a crowd sourcing system) is selected for each fragment on the basis of the errors detected in the particular fragment from a set of data cleaning assets (207) and the data cleaning assets provide a set of candidate corrections for detected data violations and then a selected candidate correction is used to replace the data in the original database.
| Original language | English |
|---|---|
| Patent number | GB2502768 |
| IPC | G06F 17/ 30 A I |
| Priority date | 12/04/12 |
| Publication status | Published - 11 Dec 2013 |