| Popis: |
In data warehousing and integration, data error repairs are very important. Some existing techniques tend to fix data errors using machine learning, statistics, integrity constraints, while others use multiple iterations to find data errors and a long time to repair them. Researches have shown the data cleaning accuracy is not efficient using those techniques. Thus to improve the accuracy, some works provided mechanisms employing knowledge bases (kbs) and crowdsourcing platforms, but using only single tables to achieve that. In this work a data error repair mechanism is proposed, which involves the use of a knowledge base on an entire database system having common attributes among its tables. Thorough experiments showed the effectiveness of the mechanism, with results showing the proposed technique is more efficient and effective using the whole database as compared to data repairs done on single tables. |
