Research On Large-scale Member Generation Method For Crowd Network Simulations

Crowd phenomena are widespread in human society.With the advent of the Internet era,crowd phenomena own the characteristics of foresight,macroscopy,long term,heterogeneity,infinity of time and space,multiple connectivity and emergence,which makes the relevant research results difficult to be observed in reality.Therefore,researches on crowd phenomena cannot follow traditional ways.Because simulation is scientific and efficient,and can simulate and verify experimental data,crowd network simulations have become one of the more effective means to support the research of crowd phenomena.Members are the basis for the promotion of crowd network simulations,which reflect individuals in reality.However,members own the characteristics of heterogeneity and large scale,which makes it costly to collect large-scale members in reality.How to generate tremendous crowd equivalent simulation members according to real members to support crowd network simulations is an urgent problem to be solved.This paper studies the generation of crowd network simulation members.The main research contents are as follows.On the premise of certain similarity of crowd networks,this paper derives crowd network equivalence according to the quality-time model of measuring crowd intelligence level,and takes crowd intelligence level as the evaluation standard of crowd network equivalence.Capability vector and task vector are defined to express crowd intelligence level.According to the feedback model of crowd network system on task completion,the calculation of crowd intelligence level is given.And the proposed crowd network equivalence has been verified by experiments to ensure its rationality.This paper proposes a large-scale simulation members generation method based on crowd network equivalence,which generates large scale groups according to small scale groups on the basis of crowd network equivalence.This method includes member metamodel definition,member model definition,crowd network structure definition and generation and member generation.Where,member metamodel definition provides patterns and member data sources for member generation.Member model definition defines types,quantities and attributes of members to be generated.Crowd network structure definition and generation generates a large-scale crowd network being similar to an existing network according to numerical characteristics of this network.Member generation generates the types and attributes of members according to the required data,and constructs the interconnection of members,which are regarded as the social relationships of real members.This paper develops a simulation member generation tool to support the method of large-scale simulation member generation proposed in this paper,to realize functions of member metamodel definition,member model definition,crowd network structure definition and member generation,and to write the generated members and their connections in XML files.