Research On Technology Of Training Network Opinion Leaders

As the speedy development of network applications, users'public opinion has a major impact on people's production and daily life, even seriously affect the security and stability of society. Because network media environment requires that people can receive the truth and remain rational, we need to guide the public opinion properly. This paper discovers the opinion leaders through speech that Internet users published and network behavior statistical characteristics of Internet users. It provides key technology to reasonable public opinion guide.This paper illustrates the purpose and significance of filtering network opinion leaders, gives classification rules of network opinion leaders clearly, introduces evaluation methods of traditional opinion leaders, and analyzes pros and cons of the traditional evaluation method for filtering network opinion leaders. It also introduces other related techniques of cultivating network opinion leaders.This paper presents the network model for screening opinion leaders, which is based on complementary characteristics of the statistical properties of the content that Internet users post on the forum and the network behavior of Internet users. Through in-depth analysis of the advantages and disadvantages in the application of online community content-based "influence diffusion model", this paper proposed a "ladder" algorithm of the evaluation method, which made the content-based evaluation method more scientific and reasonable. In this paper, the property of the evaluation method which is based on the statistical properties features is redefined, a more reasonable network statistics data is selected, and the property value of complementary is increased, in order to improve the evaluation methods. Based on the above two, the article also gives the Comprehensive Evaluation Method of the network opinion leaders.In this paper, the proposed model is realized through experiment, and the results are analyzed in depth. The method has achieved better results compared with the method that experts selecte opinion leaders artificially. The success rate of the model to select opinion leaders is more than 85%, so it has practical and important application value.