Keyword search in relational databases

As most of end users are not aware of database structure and want to make query without using SQL, keyword search in relational databases have been studied in the literature extensively. The common idea of the existing studies is to search keywords in text areas of the databases. These studies focused on two main points. First, the relation between tables should be well defined and the second, the results should be sorted in logical manner. However, run time performance of these systems is overlooked by the existing studies. In addition, current studies are developed for specific database schemas and they could not be extended for general purpose databases. Although this appears to be disadvantage, it is a necessity. In this study, we show that primary key, foreign key relation is not enough to construct relation between tables. In addition to this relation we also define new relations by using other fields which hold unique data like e-mail address or identity number. Performance analysis shows that, by using these newly introduced relations, query results are enriched.Key Words : Keyword search, Relational database