Simulation in the big data era

The big data era bring the confusions, challenges and opportunities to the modeling and simulation field tightly associated with big data. The Chinese Association for System Simulation undertook the 81st new ideas and new theories academic salon of China Association for Science and Technology. This salon, directed by Li Bohu (academician of Chinese Academy of Science) and Hu Xiaofeng (professor of National Defense University, PLA) as the leading scientists, called about 20 specialists and scholars from all the country together. They deeply investigated big data from different sides such as production, learning, research, and application so on, and A series of valuable research results were obtained.

This paper, named as Modeling and Simulation in the Big Data Era, was published in the fifth issue of SCIENTIA SINICA Informationis in 2014(in chinese). As the corresponding author, Prof. Hu Xiaofeng, who was the leading scientist of this academic salon, wrote this paper, which summarized the main viewpoints of the experts attending the salon and the main research achievements.

Big data has brought novel research thinking and methods. As we know, big data embodies many characteristics, such as networked big data as analysis object, the diversity and imprecision of data, emphasizing on correlation rather than causality, and emphasizing on in-depth analysis and indirect analysis, which make big data become a new research paradigm. However, there are different views on the question if the big data is "the fourth paradigm" or not. Experts supporting the fourth paradigm believe that data-intensive scientific research paradigm will have a significant impact on modeling and simulation. But, there are experts argue that, the paradigm transition means that a set of scientific research methods and ideas is replaced with another set of methods and ideas, and the big data is the lack of supporting on systematic theory study, technical methods and application practice, therefore big data are not powerful enough to become an independent scientific research paradigm. At the same time, others believe that, the causes of big data becoming the scientific paradigm are not enough, because of the universality of its results, limited objectivity of association rules, the ambiguity of theoretical basis on big data.

Big data brings great challenge to the modeling and simulation. The Changing big data makes modeling and simulation to face and deal with the more and more problem. The first problem is on the basic simulation theory. The simulation paradigm basing on the model, depends on the reductionism and causality, and predefines many concepts such as the target, boundary, entity, attribute, status, constraints and so on. Accordingly, it is hard to meet the demand of processing the big data on human social activities from the Internet, and it is difficult to solve the problem on disorganized world. The second problem is on the modeling method. Big data, which provides a new way how to use "data model", can establish the "data model" to understand problems, and accordingly solve complex problems behaving high complexity, with large amount of computation, being difficult to model feasibly, and create new types of model lastly. The last problem is on simulation engineering and science. It is necessary to carry out exploratory research combining of simulation-based engineering and science (SBE&S), to develop simulation paradigm, to realize the integration of the intensive computation and intensive data, to realize the unorganized causal law found on complex systems, and to achieve the corresponding SBE&S technology in big data environment at the last.

Big data also brings opportunities for modeling and simulation. Firstly, big data provides better means than ever to the simulation results analysis. The traditional simulation results analysis mostly is more direct and simple, and the big data can provide more in-depth analysis and processing in advance, and then provide a new method to solve the simulation problem on large-scale data processing. Secondly, big data creates a new way on modeling and simulating complex system. Big data focuses on the relationship, abandoning the pursuit of causality, and emphasizes the analysis of "integral data", giving up the decomposition modeling research of reductionism. Meanwhile, big data can reassemble the disintegrated fragments into the network, making us to care about the whole other than locality once again. Big data extends the "plausible reasoning" logical method, which is used to understand the objective world and has the extensive and novel application. Namely, we can speculate some results, by virtue of the existing facts, the correct conclusion, the results of the experiment and practice, experience and intuition so on. The reasoning process will be more competent in the simulation of complex systems than ever. Thirdly, big data brings a historic opportunity for development to modeling and simulation of science, including the innovation of existing simulation scientific thinking way and scientific research mode, innovation of the existing modeling methodology and simulation support system of systems, and also provides the new research opportunities to Cyber Space, intelligent simulation applications.

The discussion results of this salon will actively promote the development of simulation science and technology in the era of big data, and will make new contributions to promote the research, application, industry and the personnel training of theory and technology on modeling and simulation.

More about: | |