Schema-Matching with Data Dictionaries

摘要

We describe an algorithm for detecting semantically equivalent metadata across namespaces instantiated as database schema, an operation otherwise known as schema-matching. Assuming a metadata description discipline which imposes graph-theoretic constraints on data dictionaries, the algorithm employs analytical techniques used in information retrieval, information theory, and computational linguistics. It exploits the information inherent in textual metadata description and metadata dependency relations in order to match elements across schema boundaries.