[Purpose/significance] The Public opinion group polarization phenomenon is studied from the perspective of quantitative, to establish the quantitative calculation and polarization trend prediction method.[Method/process] Inspired from the idea of Physical fields, the interaction between the users nodes in the public opinion field is described by introducing the view potential field, and an opinion potential field intensity calculation model and opinion evolvement model of the individual power and public opinion is proposed. The natural community feature of actual network public opinion transmission, influence feature of node opinion, inertial feature of opinion and time-varying feature of virtual or true node is taken into account in the model. This model maintains that the netizen opinion evolvement is controlled by Internet public opinion field intensity and opinion inertial feature. Based on the simulation experiment, the model analyses the influence of the public opinion field connectivity, the node type conversion trigger threshold and the opinion potential influence coefficient on public opinion group polarization.[Result/conclusion] Simulation results show that the model agrees with the actual public opinion polarization evolution process, in which the public opinion field is used as the intermediary to present the interplay of netizen opinions. Compared with the model based on direct interaction, its complexity and implementation difficulty are greatly reduced, which is conducive to building a practical Public Opinion Monitoring System.