Entity Identification in Database Integration

The objective of ent i ty identification i s t o determine the correspondence between object instances f r o m more than one database. This paper ezamines the problem at the instance level assuming that schema level heterogeneity has been resolved a priori . Soundness and completeness are defined as the desired properties of any ent i ty identification technique. To achieve soundness, a set of ident i ty and distinctness rules are established for enti t ies in the integrated world. W e propose the use of eztended key, which i s the union of keys (and possibly other attributes) f r o m the relations t o be matched, and i t s corresponding ident i ty rule, t o determine the equivalence between tuples f r o m relations which m a y not share any common key. Instance level funct ional dependencies (ILFD), a f o r m of semantic constraint information about the real-world entities, are used t o derive the missing eztended key attribute values of a tuple.