【24h】

Making More Out of an Inconsistent Database

机译:利用不一致的数据库做更多事情

获取原文
获取原文并翻译 | 示例

摘要

Repairing a database means making the database consistent by applying changes that are as small as possible. Nearly all approaches to repairing have assumed deletions and insertions of entire tuples as basic repair primitives. A negative effect of deletions is that when a tuple is deleted because it contains an error, the correct values contained in that tuple are also lost. It can be semantically more meaningful to update erroneous values in place, called update-based repairing. We prove that a previously proposed approach to update-based repairing leads to intractability. Nevertheless, we also show that the complexity decreases under the rather plausible assumption that database errors are mutually independent. An inconsistent database can generally be repaired in many ways. The consistent answer to a query on a database is usually defined as the intersection of the answers to the query on all repaired versions of the database. We propose an alternative semantics, defining the consistent answer as being maximal homomorphic to the answers on all repairs. This new semantics always produces more informative answers and ensures closure of conjunctive queries under composition.
机译:修复数据库意味着通过应用尽可能小的更改来使数据库保持一致。几乎所有修复方法都将整个元组的删除和插入假定为基本修复原语。删除的负面影响是,当一个元组由于包含错误而被删除时,该元组中包含的正确值也会丢失。在适当的地方更新错误值在语义上更有意义,这称为基于更新的修复。我们证明,以前提出的基于更新的修复方法会导致棘手。尽管如此,我们还表明,在数据库错误相互独立的合理假设下,复杂度降低了。不一致的数据库通常可以通过多种方式修复。数据库中查询的一致答案通常定义为数据库所有已修复版本上查询的答案的交集。我们提出了另一种语义,将一致的答案定义为与所有维修的答案最大同态。这种新的语义总是产生更多的信息性答案,并确保在组合下关闭联合查询。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号