Keyword Query On Relational Database Based On Data Graph

With the formation of social information era, the scale of data stored in relational databasehas been in a rapid growth. How to operate the data resource is a critical task. The method ofkeyword query on relational database becomes more and more meaningful. Because the realvalue of the data is the relationship between them. As relational database is a structured data,querying on it need use complex SQL statements which make users perform querying operationsvery difficult, especially for non-professional users. Therefore, method of keyword query onrelational database has been proposed. In this method, users can query relational database withoutknowledge of complex query language and underlying database schema. Query result can reflectnot only tuple units containing keywords, but also the semantic relationship among the tuple units.Keyword query on relational database is an important breakthrough in the information retrievalfield.In this paper, further studies on the keyword query on relational database have been done. Anew solution is proposed for the problem of query efficiency and the ranking method is improved.Related studying work includes the following aspects:(1) For the problem of query efficiency, there are many studies at home and abroad.However, few scholars put parallel idea into keyword querying. So, parallel algorithm ofkeyword query on relational database based on data graph is proposed. First, convert data inrelational database into data graph. Then establish two kinds of index: keyword index and pathindex. Keyword query is achieved by traversing path index. Feasibility and how to parallel inbuilding path indexes and generating results are discussed.(2) Result includes structure and text content. Most of the existing ranking algorithm onlyconsider structure, ignore the text content. For this problem, ranking algorithm combined withmultiple influence factors is proposed. In the result ranking function, result structure isconsidered on the one hand. On the other hand, IR score of result text information is calculated.Term Frequency-Inverse Document Frequency algorithm is used to evaluate the correlationbetween result text information and keywords.(3) In this paper, use r-cliques algorithm as study background. Use C++programminglanguage to implement the proposed query algorithm and result ranking algorithm. And carry outa lot of experiments. Give a comparison and analysis of the experimental results. Theexperimental results show that the proposed method is effective.