Abstract
SQL query recommendation suggests an SQL statement to a user, based on his submitted requests and on queries of other users stored in a log. Such methods need to be scalable and data-aware. Data awareness means that the filtering condition, the most crucial element of the recommendation, contains actual values. Otherwise, the query is not directly executable. Existing approaches do not satisfy the above requirements or are limited regarding the query types supported. We in turn propose DASQR, a data-aware and scalable SQL query recommender, which also outperforms competitors regarding quality and runtimes. For the evaluation, existing approaches have proposed adaptations of metrics such as precision or recall for the SQL domain, but then only use their measures. Our comparison is broader, including new adaptations of those measures and also several existing ones.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have