Yin Zhi, Executive Vice President of Institute for China Sustainable Urbanization, Tsinghua University
The most significant change in complex systems we now live with and the history we've gone through is the emergence of a series of new tools as well as their underlying technical requirements. In the face of the future and for the sake of planning and governance, we need to build a "meta-tool" like Digital China, digital industries and digital economy that are familiar to us.This is also a consensus in the world.
The multi-source government and social big data is increasingly abundant, involving mobile communication data, Internet data, social media data, smart card swiping data, IoT sensing data, navigation and positioning data, and earth observation data, etc.The effective integration and analysis of multi-source data are playing an increasingly important role in economic and social governance.
Processing means of data technology has advanced constantly. With the support of big data, a series of intelligent analysis technologies represented by machine learning are becoming a new paradigm in the research of complex systems. They are suitable for describing complex and changeable nonlinear relations in the real world, greatly improving the capability of identification and prediction. The increasing availability of data processing tools will lead to data-intensive scientific discoveries, enabling us to describe system changes more effectively, to distinguish between rapid and high-frequency changes and medium & long term slow changes of systems more effectively, and to have better options in managing work priorities.
In addition to the improvement of algorithms, the capability of scenario design and demand scenario construction has also seen progress.Professional managers and scientists of certain domains should be project planners and project directors in IT companies, so that the companies are able to be more proactively combined with real scenarios of management, operations and construction, rather than just for the sake of algorithms. Technical tools alone are not enough to solve problems. It is necessary to effectively coordinate with application patterns.
The algorithms of big data can solve data sources technically, and discipline groups can generally solve the issues of model selection, model maintenance and feedback & verification.However, from the perspective of application patterns, data collection and definition of cases are more important. In data collection, we need to clarify data logic rather than simply solve the issue of data sources.Model selection is based on the definition of cases, and a wrong definition will lead to problems with the model. It requires effective coordination of various application models. And ultimately, we will see the effective connection of sensing, measurement, simulation and mining that data technology is capable of with observation, action, adjustment and decision that we need in real social management, so as to make progress by turning technology into scenario-based things.
The research on planning, decision-making and management optimization will apply to comprehensive or dedicated spatial governance. Seen from current situation, we have made progress in some issues, e.g., multi-source fusion of government data and social data, the single-point modeling & analysis of traditional econometric models as well as typical application, data-driven machine learning, deep learning and reinforcement learning, and further promotion of typical applications based on scenario analysis.
Published on June 23, 2022