Download now Free registration required
In this paper the authors propose and experimentally evaluate a data integration approach where the uncertainty generated during the comparison and merging of the input data sources is included into the resulting mediated schema, and can be used to provide richer answers to the users. They describe a system implementing the method, and use it to empirically study the impact of uncertainty management on the effectiveness and efficiency of the data integration process. In particular, they test the approach on benchmark datasets, showing that considering uncertainty they may increase the recall of the method, and on real databases, showing that it can be applied to large data sources.
- Format: PDF
- Size: 380.8 KB