Relational Database Keyword Query Processing Key Technology Research

With the fast development of database technology, keyword search over relational databasesenables casual users to use keyword to search relational databases just like searching the Web,without understanding the underlying database schema and complex SQL query language. Currently,the keyword search system of relational database still has some problems. For instance, theefficiency of query system is not high; the accuracy is low, without supports for phrase query; andthe result of query is quite simple. Addressing the above defects, the paper mainly proposes someways for improvement.Firstly, we propose the CNI method to improve the production rate of candidate networks. Theoriginal query system based on schema graph starts to make breadth-first searching after producingset of tuples graph. It is time-consuming. The study finds that, aiming at OR semantic query, we cantake advantage of the method to establish candidate networks index to achieve the candidatenetworks that we need.Secondly, we propose the P-B method to effectively indentify the phrases in the relationaldatabases and provide the feedback of phrases we recognized to the users. Then the users choose thephrases which is qualified for the query, so as to improve the query capability and the queryaccuracy.Thirdly, we propose the TCD method to cluster the query result joinning tree with effect. Itclassifies the joinning tree with the same semantic into one group and describes each cluster, whichis convenient for users to understand the semantic of query results and rapidly seek out the accuratequery results.Theoretical analysis and experiments show that the above three methods are effective.