Research On Opinion Spreading And Consensus Of Complex Networks

In real word, it is an important aspect of social behavior dynamics that opinion or decisionchanges and reach consensus. The study of evolution and formation process of opinion helps revealthe nature of the public opinion. And it is of great significance in explaining and understanding theevolution and development of humans’ complex group behavior. We summarize the contributionsof this paper as following three points:1. In the process of opinion spreading, the node which is an active opinion spreader might be animmune, or continue to remain a spreader. Based on this idea, this paper proposes a novel opinionspreading model, which is based on SIR model. We separately perform experiments on the WS, BAand Facebook networks. We obtain the experimental conclusions that: the opinion of source nodewith high degree is propagating faster than that of source node with low degree, the Facebooknetwork needs more spreading time(larger than those of WS and BA networks) to transmit theopinion of source node to other nodes, the Facebook network does not exist an spreading thresholdin the opinion spreading.2. In the process of public opinion spreading, the opinion will become more briefer, moreconvenient to be obtained. Based on this idea, we propose an opinion spreading model, which takesopinion revision into consideration. We believe that the opinion of one node has some probability tobe modified. The experimental results indicate that: whether the nodes’ opinion of a network (WSor BA network) at the final state relate to source node’s opinion or not depends on the networknodes’ opinion revision rate and the opinion strength of source node. Besides, for the Facebooknetwork, this not only depends on the above two factors, but also depends on degree of source node.3. In consideration of the facts that, in the process of opinion spreading, nodes prefer to selectcertain nodes to communicate and they have memory for viewpoints during the communication, weestablish a novel opinion dynamics model by extending the Deffuant model. Priority selectionstrategy and the memory effect of node are adopted in our model. And we study the influences ofthese two factors on network opinion formation. The experimental results indicate that the proposedmodel adopting priority selection strategy helps decreasing the time of consensus in non-uniformnetwork. And the joining of the memory effect can not only promote formation of networkconsensus, but also make the network reach consensus at a small threshold.