Learning database queries via intelligent semiotic machines

In the Big Data era, data-intensive and data-driven approaches are affording the paradigm change. In this data deluge, finding relevant and appropriate information is of great importance for achieving the demanding goals posed by business and society. However, traditional database search is limited by the assumption of query criteria as hard constraints. This provides results without personalization and flexibility, and often leads to information overload, i.e., non-focused results that are not informative to users. There is also often relevance misalignment, which occurs when the system only relies on consensus relevance. To overcome these problems, we present, in this work, a novel approach for (semi) automatically learning database queries based on Semiotics and Computational Intelligence techniques, in which user perception and intention are considered for interactively producing tailored queries that facilitate personalized data exploration and retrieval. In this sense, we present an empirical analysis to assess the effectiveness of our approach in learning SQL queries in a Query By Example scenario. The obtained results confirm that our approach is capable of generating SQL queries based on a single user-defined example without requiring any database-specific knowledge such as query language or database schema and structure.