On the Usefulness of SQL-Query-Similarity Measures to Find User Interests