Research On Technologies Of Keyword Search Over Relational Databases

Traditionally, SQL is the main interface to access data from relational databases. However, it is difficult for inexperienced end users to learn the complicate syntax of SQL. Enabling keyword-based information retrieval over relational databases will allow users to acquire information from databases without any knowledge of SQL and underlying database schema, just like the way of common search engines. Keyword Search over Relational Database (KSORD) techniques have been focused in the current database field.This paper researches the methodes and the key technologies of KSORD such as isystem framework, system model, data model, query language, Top-k ranking and query result for committing. The work presents formal definition of KSORD based on relational database management by'analysing the implementation strategies of relative systems. Basing on the integrity constraints, hypothesis and the question model, the work creates the data model and the query method, so the system frameword of KSORD is created.This paper designes a new query Syntax, the distinguish decomposition algorithm by analysing the systaxes of prior systems and presents a way for searching metadata. According to the kinds of the relation attributes in relational database management, the work analyses the equivalence value queries and the fuzzy value queries of the numeric attribute and the1 non-nuberic attribute, and presents the tuple scoring algorithm of the equivalence value queries for the numeric attribute and the realization way of fuzzy queries based on Rocchio algorithm. The scoring table and the figure of the scoring table are created to expand the candidate tuple join tree. The paper presents a new sorting method for using non-monotonous functions based on the virtual document model by analysing prior sorting ways and principles. The method considers serveral aspects such as integrities of other results and classification factors, it uses regulation parameters for adding some relational Operators such as AND and OR into the sorting model. The work modifies the prior sorting methods to a new sorting method by using join different tuples. The non-monotonous of the sorting function improves the shortage of prior methods, which cannot apply practice, it reduces unnecessary database accesses to enhance the query efficiency.This paper realizes a system of KSORD, which employys multilayer to realize keyword search. The work presents the experimental anlaysis for the influence factors of the numeric attribute, the non-numeric attribute and Top-k queries, and the main influence factors of the fuzzy queries of the two kinds of attributes. The results show that the relational methods of the numeric attribute and the non-numeric attribute is effective on system overload and efficiency.