Recommending Join Queries Based on Path Frequency

Real databases often consist of hundreds of innerlinked tables, which makes posing a complex join query a really hard task for common users. Join query recommendation is an effective technique to help users formulate better join queries and explore their information demand. In this paper, we propose a novel approach to automatically create join query recommendations based on path frequency. Our approach generates recommendations by analyzing the database schema and underlying data. First, we exploit join queries which are likely to be queried by considering both the importance and the connectivity of tables. Second, we provide users two recommendation forms. One needs no input information and the other allows users to input incomplete information. Users can choose one according to their knowledge. Extensive evaluations demonstrate the effectiveness of our approach and show that our method is helpful to formulate good join queries in practice.