Detecting Irrelevant Updates Using Additional Information

We address the problem of determining whether an update is irrelevant to a view, i.e. whether a view is changed by an update to one of its source relations. This problem has application to areas such as query optimization, data warehousing, integrity enforcement and data integration. The class of views we consider are conjunctive queries with equality constraints and the updates considered are insertions, deletions and modiications. The signiicant novel aspect to our approach is that, unlike previous work on the problem, we assume that the view instance is available in addition to the view deenition and the update in detecting irrelevance. This allows a wider class of irrelevant updates to be detected than by using the view deenition and update alone. In our approach, the test for irrelevance consists of two phases. In the rst phase, a test in the form of a parameterised query is generated at view deenition. Then at update time, the values of the view and the update are substituted in the test and executed. This approach has eeciency advantages over other approaches where all processing is done at run time.