New Publication: Towards Mega-Modeling: A Walk through Data (Sigmod Records)
This paper reflects one result of Prof. Freytag's cooperation with Italian research partners during his sabbatical in 2012. Reference: SIGMOD Records 42, 3 (September 2013)
Big data is perceived as a fundamental ingredient for fostering the progress of science in a variety of disciplines. However, we believe that the current ICT solutions are not adequate for this challenge. Abstractions and languages for big data management are tailored to vertical domains and influenced by underlying ICT platforms, hence unsuitable for supporting “computational interdisciplinarity”, as it is required if one wants to use the best of, e.g., analytical, inductive, and simulation techniques, all at work on the same data. In other words, “our society is data-rich, but it lacks the conceptual tools to handle it”.
In this paper, we provide some evidence that megamodeling is a viable approach to data analysis by using a bottom-up, inductive method. We consider several experiences of data analysis research performed at our home institutions and examine them in retrospective, inducing their mega-modularization a-posteriori. This exercise convinces us that the mega-modeling Approach could be highly beneficial.