Prezentujący: Denis Martins (doktorant), University of Münster, Germany

Abstract: Formulating database queries in terms of SQL is often a challenge for journalists, business administrators, biologists, healthcare professionals, and the growing number of non-database experts that are required to access and explore data. Moreover, writing accurate SQL queries is especially problematic when users lack sufficient knowledge about either the query language or the data domain. In these situations, query formulation becomes a highly interactive, time-consuming process and yields results that frequently do not meet user preferences and information needs. These problems have motivated the development of Query By Example (QBE) systems in which users are not required to possess any database-specific knowledge nor programming skills. Instead, users are asked to provide a set of data examples that satisfy their mental query and the QBE system tries to (semi-)automatically construct (i.e., search for) an accurate query, if one exists. In this talk, we review state-of-the-art QBE approaches and describe their advantages and limitations. We also demonstrate how machine learning algorithms can be applied in the context of QBE and show the effectiveness of three specific algorithms, namely, Greedy Search, Genetic Programming, and CART decision trees in learning queries in two distinct databases. Finally, we provide a research agenda describing future research development on QBE in the context of the democratization of data retrieval and exploration.

Short bio: Denis Martins is a Ph.D. candidate at the Databases and Information Systems group of the University of Münster, Germany. His research interests focus on the application of Computational Intelligence and Computational Semiotics for improving the democratization of data retrieval in Big Data scenarios. He holds a Master’s degree in Computer Engineering from the University of Pernambuco, Brazil, where he researched hybrid computational approaches for intelligent decision-making support.