Research On User Follow And Information Retweet Prediction In Enterprise Micro-Blogging

As the "Internet +" action plan is put forward, the Internet will realize the deep integration with the traditional industry, will greatly promote the development of electronic commerce. With the rapid development and wide application of micro-blogging service, the advantages of which in terms of communication and dissemination of information have become increasingly prominent. As of December 2014, the scale of China’s microblogging users has totaled 249 million and the Internet users’ utilization rate reached 38.4%. At present, microblogging has become an important marketing tool for enterprises. According to CNNIC survey, among all the Internet marketing tools used in business, instant messaging, search engines and e-commerce platform rank the top three, the utilization rate are 62.7%, 53.7% and 45.5% respectively. Microblogging marketing ranks the seventh, with the utilization rate of 26.9%, with an increase of 6.2 percent compared with the figure in 2013, showing a greater demand of enterprise for microblogging marketing.However, compared to the increased enterprises using microblogging marketing, the size of microblogging users has reduced 31.94 million in comparison to the scale at the end of 2013. Internet users’ utilization ratehas also decreased over the previous year by 7.1%. This trend is also affecting users’ follow and use of the enterprise microblogs. Therefore, how to attract more active users to followenterprise microblogs and to be actively involved in the dissemination of microblogging information has become a common issue for the industry and academia that demand surgent studies. To solve the small size of active users and the narrow scope of information dissemination of enterprise microblogs, this paper has conducted exploratory research and the main contents include:(1) Establishing the model of user’s follow intentionBased on relationship marketing perspective, the model regarding user’s follow intention for enterprise microblogs has been built and the five key factors affecting user’s attention on the enterprise microblogs have been brought up. Empirical analysis using structural equation model has been conducted. The empirical results show that there is a positive impact of satisfaction on follow intention; a significantly positive impact of satisfaction on trust and corporation image; there is also a positive impact of communication benefits and quality of information on user’s satisfaction. In addition, there is some mediating effect of satisfaction on the relationship between corporate image and follow intention. The findings make clear the importance of various factors and provide a reference for the enterprise in taking appropriate measures.(2) Constructing prediction model of the follow behaviors of micro-blogging usersThe user follow predictive index has been established in accordance with the formation mechanism and influencing factors of the user’s follow intention together with the influence of microbloggers, user’s activity and interest similarity. The fuzzy proximal support vector machine has been proposed based on density-weighted method, the fuzzy membership function optimized and the user follow predictive model established. The empirical research has been carried out by adopting the open data of Tencent microblogs. The contrast test has been conducted on a variety of support vector machine model and a variety of fuzzy membership degree algorithm in the experiments. The results show that the density-weighted fuzzy proximal support vector machine obtains the best average prediction accuracy. The research conclusion provides an effective method and a tool for finding out active users of enterprise microblogs.(3) Building the intention model of microblogging information disseminationCombined with personal motivation, the research model regarding the intention of enterprise microblogging information dissemination has been proposed by drawing lessons from the relationship marketing theory to investigate the main motivation and influencing factors of user’s participation in the dissemination of microblogging information. The conclusion of empirical analysis shows that the relationship marketing theory can be applied to the research of the dissemination mechanism of microblogging information. There is a significantly positive impact of satisfaction, trust and commitment on the intention of information dissemination; satisfaction has a positive impact on the trust and commitment and trust has a positive impact on commitment. Meanwhile, the empirical analysis has also verified the positive impact of prestige and altruism on the information dissemination. On the analysis of satisfaction impact, the empirical results show that the quality of information and benefits of communication have a significantly positive impact on user’s satisfaction.The mediating role of commitment and trust on the relationship between satisfaction and intention of information dissemination has been examined through regression analysis. The empirical analysis shows that there is partial mediating effect of the commitment and trust. The test of moderating effect verifies the similarity of interest generates moderating effect on the relationship between satisfaction, trust and commitment, the three variables and the intention of information dissemination; there is also a mediating effect of interactive tendency on the relationship between prestige and intention of information dissemination.(4) Constructing the predictive model of microblogging information retweetingBased on the studies of influencing factors regarding user information dissemination, user trust network has been built by taking advantage of user’s social behaviors; the algorithm model with random walk of trust has been improved by introducing user trust and interest similarity and the predictive model on the microblogging information retweeting has also been established. The prediction of information retweeting has been conducted by extracting data from enterprise microblogging platform. Experimental results show that the proposed model has preferable prediction accuracy and recall rate and the model further validates the user interest modeling and trust calculation model have preferable applicability in the prediction of information retweeting. The research conclusion provides a better method and tool in searching active information disseminators of enterprise microblogs.The conclusions obtained from the studies of user follow and information dissemination of microblogs play a positive role in enriching the studies of relationship marketing theory in the social platform, especially microblogging network and has important practical significance and promotion for enterprises finding out active users and accurately disseminating marketing messages.