The SimilarQL framework: similarity queries in plain SQL

As the variety and complexity of collected data increases, also does the need to analyze them by similarity. However, current Database Management Systems (DBMS) do not provide effective support for similarity queries, and the research on the subject has, until now, provided only a limited support through tools that are not simple to be deployed nor maintained. Thus, what to do if you need to perform a quick exploratory analysis over your data, and must do it 'now'? In this paper we show how to use the readily available support from standard DBMS to target this issue, and present the simple, yet powerful SimilarQL framework, which provides a complete and flexible set of similarity query operators. It can be readily deployed over conventional DBMS, without depending on the bulk data structures and software systems required to handle long-term standing similarity queries. We present results from a real-world dataset regarding the query execution times using SimilarQL, and show that acceptable runtimes may be achieved, while intuitively and easily exploring complex data by similarity.