Research On A Database Schema Matching Validation Method

At present,such as database schema improvement or ingoing, the development based on the componet, database design etc, a large number of database applications come with the tide of fashion, in which schema information plays a key role, and people will face the problem that how to operate the schema information. Schema matching operation is considered as the best way. As the rapid growth of the network databases and E-business integration, research on the schema matching is motivated further. For the generated results of schema matching, users can't be aware of them correct or not, or whether they conform to the expectations and requirements, and the schema validation technologies are required to aid people in making the decisions.After deeply studying and analyzing the current database schema verification field, some schema verification methods have some deficiencies that generality is not strong, or they are only applied to schema mapping validation process. To address this issue, this paper proposes the research on a database schema matching validation method, and the core of the method is DSMV verification algorithm, namely the set of the candidate attribute matchings is generated by means of one provided by human or existing methods, then the algorithm ordinally calculates the similarity of the attributes in candidate attribute matchings by using the constructed decision tree. If there are the same similarities, one of them is retained by using the method on the Functional Dependencies. For the candidate attribute matchings of which these similarities are greater than the threshold, their closures are calculated, then some candidate attribute matchings may be retained by using the generated closure sets. At last, the verified attribute matchings are provided to users.The DSMV verification algorithm is verified and analyzed by experiments. Experimental results show that DSMV algorithm has the higher precision rate and recall rate.