Multi-agent Based Problem-solve Model And Simulation In Mass Collaboration

With the development of the internet and web2.0, costs of distant communicationhave been substantially reduced, which enables people all around the world to conducteffective collaborations. Thus, Mass Collaboration came into being. Meanwhile,global issues in environment, society and economy brought by globalization requiresfurther collaboration among all peoples regardless of regional restrictions. As a result,the conception of Mass Collaboration has been widely concerned since its appearance.The thesis thinks that problem distribution and solution are extremely important in aneffective mass collaboration. So we try to model the problem solution of masscollaboration and then simulate it in a multi-agent way.The thesis starts with collaborative problems and the participants’ expressions.Relying on the computer color display concept RGB, we project expressions andcollaborative problems that cannot be directly measured as well as the knowledgeattribution of participants into the RGB color space, then turn them into visiblefigures, making it possible to calculate the distance between the knowledgeeigenvalues of different individuals. In the thesis, we divided different knowledgeattributions of participants into three types, they are homogeneous knowledge type,consistent similarity type and factors similarity type. Based on this, we built ageometric space model of participants and problems. By analyzing the nature of tasks,we set up3rules: knowledge similarity rule, masters’ rule and mass-election rule, andthe corresponding model of solutions. Then, we created a multi-agent simulationsystem for mass collaboration by adopting the Repast Simphony2.0. In this system,we define the Agent objects and their network structure for participants and problemdistribution center. At last, by comparing the simulation results, we analyzed theeffect and efficiency of task solutions in different models.