Automatic user steering for interactive data exploration

The amount of data that have flooded databases during the last few years have created several new problems for the data management community to address. One of the most prominent is the discovery of new and interesting information that is hidden in the underlying big data sets. As of now, in order to explore these data sets users begin with a few general queries, study their output and iteratively issue more specific ones until they discover interesting information. This is an onerous process that requires time and effort. In this thesis we are developing an automatic data exploration framework that will make the discovery of new information in a vast data space both effective and efficient. Our system asks users for their relevance feedback on strategically collected samples in an interactive manner, steers them towards the interesting parts of the database and eventually formulates the query that retrieves their data of interest. Our preliminary results are encouraging and allow us to persevere in the development of such a system.