Estimating the Quality of Data in Relational Databases.

摘要: With more and electronic information sources becoming widely available, the issue of quality these, often-competing, has become germane. We propose a standard for rating with respect to their quality. An important consideration is that often varies considerably when specific areas within these are considered. This implies assignment single an source usually unsatisfactory. Of course, user overall may not be as this extracting from source. Therefore, methods must developed will derive reliable estimates provided users, specifications have been assigned sources. Our work here bears on all concerns. describe approach uses dual measures gauge distance in database truth. then combine manual verification statistical arrive at useful databases. consider variance by isolating databases homogeneous quality, estimating each separate area. These composite regarded affixed database. Finally, we show how individual queries such specifications. was supported part DARPA grants N0014-92-J-4038 N0060-96-D-3202.