The objective is to facilitate the process of making sense of and adding value to massive and heterogeneous data, either for deriving new knowledge or for taking better actions. Nowadays, this process is semi-automatic and relies on the analyst's expertise and understanding of the domain, the (dirty) data at hand and a plethora of complex tools.

We envision a novel generation of data analysis approaches where the design of the most adequate pipeline for data exploration and analysis is automatically suggested to the analyst, tested and refined, and only the most relevant results are shown to the user.