Description

Abstract

          The current research is mostly based on hybrid approaches. Both schema and instance-level

approaches present deficiencies that may be solved when they are combined. In real-world

situations there is no information about the schema structure is available. Only the data instances

may be available for use. The instance-level approaches build their strategies to find matches

between schema elements using only the data instances available in the repositories being

processed.   The graphic tools are used to find the schema matches. But this task is usually

manually done. A novel evolutionary approach to addressing the problem of automatically finding

complex matches between schemas of semantically related data repositories is proposed. This is

the first approach that is capable of discovering complex schema matches using only the data

instances. Since we only exploit the data stored in the repositories for this task, we rely on

matching strategies that are based on record deduplication and information retrieval techniques to

find complex schema matches during the evolutionary process.