Study On Keyword Query Of Relational Database

With the development of relational database technology, More and more data isstored in the relational database. Relational database was the mainstream form ofdatabase in the current,and the queries technology for relational database is alsoincreasingly important. However, due to the structural characteristics of relationaldatabase,should use the structured query language(SQL) to query it’s data. It requireusers to learn the complex structured query language and understand the databaseschema. Obviously,this is a difficult thing for an inexperienced user. Therefore, Find anew way to query the relational database information is one of the hot areas ofresearch in recent years.In the research of keyword queries with the relational database, the core idea arebased on graph and simplified subtree mostly. According to the generation strategy ofthe simplify subtree the research of keyword query with relational database can bedivided into two categories. That is the method based on data graph and the methodbased on the schema graph。The former is processing on the data graph directly,toenumerate the simplify subtree from the data graph. And the latter used to thedatabase schema to create the join expression, then execution the join expression ofcorresponding SQL statement on the DBMS to get the simplified subtree.This paper adopt backtracking to the data graph with depth-first search method todetermine the tuple information corresponding to the keywords. In the retrievaling ofthe keywords based on the database,the solution space is defined as the graphstructure of the database and the solution space graph nodes are constituted bydatabase tuple, the main foreign key relationships between tuples constitute the edgebetween the nodes of the graph.Backtracking depth-first search solution space to get problem solutions,that is query results related to key words. In the searching process,backtracking algorithm is not preserved the completely solution space in thecomputer, but simultaneously generating nodes in the answer tree, checking, anddiscarding the node.At any time of the search process, saving only the nodes on thepath from the start node to the current node, thereby reducing the memory usage.Defined pruning function in the node to cut nodes which are not possibly containedin the problem solution as early as possible to avoid search the unnecessary subspace,improving the search efficiency further.