Analysis On The Evolution Of Online Public Opinion Based On Social Network

Currently, public sentiments caused by Internet come out one after the other. The information on the Internet is so complex and it has concealment, divergence, and permeability and randomness characteristics. The spread and outbreak of public sentiments, which is caused by or grow from social public crisis, are irreversible trend. It’s of great significance to do research on the evolution of public sentiments in the Internet platform. More and more researchers pay attention to this field. If we treat the users of the public sentiments as notes, and treat their communications as lines, the public sentiments can treat as a social network as a whole. The evolution of the public sentiments research involves networks science, communication, and information science and so on. This paper tries to find the patterns about the evolution of public sentiments through the structure analysis of the public sentiments social network. The main and innovatory results are as follow:Firstly, on this paper, we choose the public sentiment, "Officials couple beaten nurse", as subject. And we develop a social network,in which the Weibo users as nodes and the forward relationships as lines. The key nodes are recognized with the improved PageRank method. Some patterns are found out. The nodes of media play an important role in all the periods. The nodes of the officials involve in the public sentiments social networks lately. So they only have a little effects during the periods of evolution and outbreak. While the nodes of gross roots have an underestimated influence during all the periods. So the nodes of the officials should involve in the social networks as soon as possible to improve the discovery and mastery ability of the public sentiments.Secondly, the time variable is added into the static social network. And we analyze the trend of the network and nodes measurement indicators, to try find out some patterns. The results show that there are clear phases during public sentiments.Thirdly, we do research on the evolution of public sentiments in different social networks and compare all the results. Current simulation researches mostly use the random network, scale-free network and small world network. The variables can’t be directly correspond to the real life. So all those can’t explain the exact patterns. On this paper, we use the real data from Weibo, and try to find out the effects of different initial parameters to different phases.The results show that the evolutions of the public sentiments are different in different phases. Government shall take different methods to control and manage the public sentiments.①During the first phase,the numbers of the nodes are very small. But the spread range can be very huge. So the earlier the government involves, the better the effects of the control and management. ②The size of the social networks reach the maximum during the outbreak phase. Governments shall add more key nodes.③During the recession and maintenance periods, the key nodes from governments can reduce relatively. But governments shall still keep monitoring the topics and emotions.