Research And Implementation On Keyword Search For Semantic Cloud In Relational Database

In the past several decades, the relational database system has developed rapidly, and has been applied to everywhere of the life. At the same time, the requirement of people for datamanagement, especially in datasearching, has increased. But in this respect, the relational database itself cloudn’t meet the requirement from people. However, the occurrence of the keyword search technique in relational database (RDBKWS) makes people be able to access data from the database without any knowledge of the relational database or the underlying schema of the data, as easily as using the search engine. But the traditional RDBKWS only returned the results that match the keywords, and did not pay any attention to the underlying relationship between the results.First of all, this thesis presents the definition of Semantic Cloud, in order to return the underlying semantic relationship between the RDBKWS results to users. Based on Semantic Cloud, this thesis presents the keyword search strategy for single keyword and multiple keywords separately.In the single keyword search strategy, this thesis presents the definition of semantic frame, according to which the tuple containing the keyword could generate the SFresult. The SF result is generated by linking the tuples in database according to the tuple containing the keyword and the semantic frame. Then the system processes the SFresult into Semantic Cloud and returns to user. In the multiple keywords search strategy, thie thesis presents a new type of candidate network, semantic candidate network, which could handle multiple keyword search with OR semantic. The SCN result can not satisfy all of the keywords. After that this thesis presents the Semantic Cloud generation algorithm. In the algorithm, Semantic Cloud is created by integrated the traditional RDBKWS results and returned to users.At the end, this thesis presents the special ranking strategy for Semantic Cloud, SC-TF-IDF strategy, based on TF-IDF strategy. The core idea of the algorithm is that calculate the score of Semantic Cloud by integrateing the TF-IDF score of the traditional RDBKWS results. Then this thesis improves the SC-TF-IDF strategy. The experiment confirms that the improvement raise the precision and the recall of the results.