A Specificational Approach to Merging Persistent Object Bases

We examine the problem of merging persistent object bases (POBs) and the attendant problem of determining when two objects are equivalent. In particular, two or more source POBs are combined in a single target POB in which physically diierent objects in the sources may be merged into a single object in the target. Although persistent object base research has not focused on these problems in the past, any attempt to model (portions of) the real world in an object base butts up against them. In fact, the object-centered nature of POBs is consistent with and almost motivates our two-phase solution to the problem, in which anòbject identiication' phase precedes a `constraint resolution' phase. Families of keys are used to identify objects, after which various mechanisms are used to merge the remaining data. For example, preference speciications are used to resolve constraints violated in the target { by preferring facts from one of the sources over the others.