Keyword Search Over Relational Database And Data Integration

Nowadays, it’s difficult to get data from relational databases and data integration systems without any knowledge of schema and SQL. The development of keyword search engines such as Google highly improves the ability to retrieve information. Applying the keyword search to relational databases and integration systems has been the important research area.The traditional keyword search systems over relational database are based on the full-text index which supported by Oracle, MySQL and so on. These systems cannot be applied to the databases without the index permissions, especially the data integration situation. The existing data integration approaches over relational databases do not change the pattern of retrieving information which is hard for users to use.The paper has the following contributions:1) the paper makes a conclusion about the systems and theories of the keyword search over relational database and data integration.2) The paper implements a keyword search over relational database system using the metadata which can be applied to databases without index. The system allows users mark the semantic meaning to keywords and takes samples from database to improve the mapping between keywords and database terms. The system also put forward a new method to rank the final search results which combines the keywords mapping score, the number of keywords and the number of tables in SQL together to rank the results.3) The paper implements a data integration system based on the keyword search system over relational database. The system assigns the uncertainty to the edges which cross different data sources, and it allows users give feedbacks which can help the system to build the relationships better between different data sources. It also improves the ranking method by taking the uncertainty of edges into consideration. At last, the paper does a lot of experiments to verify the systems.