| 摘要 |
<p>Disclosed is a method of estimating a probability of error of a value of a particular attribute in an aggregated database. The aggregated database being compiled from several source databases, each containing values for some of the same attributes. The method starts by estimating a probability of error for each source database, the probability of error for a source database representing the chance that values of attributes in the database are incorrect. Then the probability of error of the value of the particular attribute in the aggregated database is estimated from the estimated probabilities of error for the source databases. This is done by determining the number of conflicts between respective pairs of the source databases by comparing values of attributes in the source databases, and deriving the probability of error for the source databases from the numbers of conflicts. The estimated error probabilities may be obtained using a pari-mutuel technique.</p> |
