| Without knowing the SQL queries or database schema, casual users can use Keyword Search over Relational Database (KSORD) techniques to access databases in a fashion similar to using search engines to search the Web. KSORD techniques have been focused in the field of database. However, the key techniques only realize exact queries, not fuzzy queries. Our papers develop fuzzy queries of keywords in numeric attribute and make further study of result sets ranking, which are deeply based on researching the system named SEEKER exploited by Prof. Wang Shan .etc.When we make the fuzzy queries of numeric keywords, the main method is that we first analyze the distributing of databases, and then propose the Membership Function and the system ranged parameter. In addition, with the concept of fuzzy arithmetic operators in order to adjust uncertain range, fuzzy queries have been transmitted to near restrict exact queries. In the whole, we creatively recognize the difference value as a factor of Score Function aimed at numeric attribute. It ensures that the smaller the difference value is the frontier the tuple ranks.Casual users are the estimator. We know that the Score Function of SEEKER with ranking result sets is related to several factors such as sizeof(T), but they greatly influent accuracy of ranking in case of non-normal. In order to improve ranking performance, we treat the factors as normalization using the vector space model method. It is proved to be optimized by means of statistical analysis.Thus, the main contribution of this paper is to creatively propose Membership Function and define the fuzzy operators. Based on the above, we can realize fuzzy queries of numeric attributed keywords. What's more, our paper describes an instance to make the process clear. Meanwhile, the result sets are proved to be more superior with normal factors than without normal method. All the analysis is based on the vector space model. |
PDF Full Text Request