Top-N Keyword Search Based On Knowledge Base In Relational Databases

With the rapid development of all kinds of structured data over the web and computers, the number of users acquiring to access the queried information over relational database increase greatly. However, traditional querying methods require users to know about SQL querying languages and database schema; In order to obtain the needed results, users wish that free-form keyword search can be completed over relational database such they can avoid creating SQL querying statements and only need to submit some query keywords to the query interface.This paper presents a new method based on a knowledgebase and realize keyword search in the free form over relational database. First of all, in order to make sure space and time cost small, this paper improves and integrates the storing structure to compress the redundant information stored in the index table. In order to make sure search effectiveness, this paper employ elaborate storage granularity to store frequency of tuple words, words length for a tuple, the total numbers of tuple words and so on, and create knowledgebase for each relation in relational database. Secondly, fully considering the structures and factors in relational database, we improve classic Information Retrieval ranking strategy and assign different weight values to attributes according to the matching degree of relation and query words to satisfy different users'querying intention. Sequentially, in order to enhance keyword search efficiency, system translates the tuple set networks to SQL query statements according to sequence of attribute weights and returns the Top-N answers for users basing the different interest about the results, and then according to requirements, all ranked results may be displayed for a query. In the last experimental results on a large real dataset indicate that the performance of the method is efficient and effective.