Research On Materialized View Maintenance Method Over Keyword Search In Relational Databases

Keyword search over relational database has become a hot topic of database field. However, if each query is re-executed, that may cost too much. Materialized views can effectively improve the efficiency of query, therefore, each intermediate result and top-A: results are stored as materialized views, and maintaining keyword query result by maintaining materialized views are studied in this paper.Recent research mostly focuses on the SQL oriented maintenance of materialized view. Existing methods on incremental maintenance have the problem that the maintained result is approximate and can only make an optimization on the maintenance of single view, and the maintenance of top-k results of keyword query has not been widely studied, the efficiency of maintenance is low too. To resolve the problems above, three parts of research are conducted in this paper:The first part is about an algorithm for the generation of multiple views based on common. Common expressions are selected from candidate networks with high relevancy to keyword search, the path of query results is stored via Huffman tree, each intermediate result is stored as a materialized view, and the final query result is the root node of Huffman tree, auxiliary rows are added to the materialized view. Experiments show that the method can improve the query efficiency.The second part is about the accurate and efficient view maintenance method. According to the Huffman tree, a method of quickly updating view based on provenance is presented to modify irrelevant data from keywords database. And for inserting and deleting data, we put forward a precise method of incremental maintenance based on provenance. The experiments have shown that the two methods can reduce the cost of maintenance and improve efficiency of query, and the precision of maintenance is 100%.The third part is on top-k results of keyword search. A method of approximate quick maintaining view is presented to modify irrelevant data from keywords database. For modification related to keywords, based on the method above, top-k results are maintained. And in the process of maintaining, some optimization conditions are considered, so optimized method of incremental maintenance on top-k' is proposed. The experiments in this paper benifise that the method can effectively improve the maintenance efficiency and query efficiency of top-k results.