| With the development and integration of mobile communication technology and information network technology, Internet steps into Web3.0Era, which is also called Mobile Internet Era. This stage is "application service" oriented and characterized by mobile, ubiquitous, intelligent, personalized and diverse service. Microblog, because of realizing seamless connection between mobile terminal and internet and having advantages such as facility, free use, instant arriving, fast transmission and so on, has become an important sharing tool and self-media platform in less than three years. It is very important for related government and enterprise marketing department to understand user behavior in microblg community and make good use of the microblog platform.In traditional studies, it is usually assumed that users’ behaviors are random in time and thus can be simply described as Poisson processes. However, as the ways of data collection and capability of data processing are becoming significantly improved, more and more empirical studies on users’ behaviors prove that many behaviors deviate from the Poisson distribution. The empirical research covers almost all the traditional internet behavior, but rarely on mobile Internet. Also, the research on mechanism which drives human behavior is less. At present, the mechanisms mainly include priority choice, interest-driven, memory impact and so on. However, because of the high complexity of human behavior, single mechanism can only be used to explain certain human dynamic behavior and has no universality. From the model validation perspective, all models can generate specific range of power exponent, but lack quantitative parameters. So, it is hard to apply the model to simulate human behavior. The existing research on microblog at home and abroad focus on information transmission and the researchers are from the fields of news and communication. There are also some research from the perspective of micrblog motivation and micrblog marketing. Research methods are usually qualitative analysis like questionnaire, case analysis. Only a little analysis about the user behavior in microblog community based on the real data of the website, including frequency analysis, correlation analysis and other basic statistics. Although most foreign research combines quantitative and qualitative analysis, it should to be strengthened in depth.Considering the blank and insufficiency of existing research, the study object is microblog users from weibo.com, which is the most popular microblog website in China. The paper focuses on users’posting, forwarding and comment behavior and takes human dynamic theory as theoretical basis, adding with complex networks, statistical physics, probability, statistics and management science theory.The main research conclusions are as follows:Firstly, the paper focuses on posting behavior and presents a human dynamic model co-driven by interest and social concern. Statistical result shows that users send messages in burst and periodically in the microblog community. Information release peaks around12am both at weekends and on weekdays, and that is quite different from the peak of using email, instant communication devices, and mobile phones, which is usually at10am. This lagging peak of microblog indicates that most messages on microblog community have nothing to do with work, and are just small amounts of talk in fragment time.Early in microblog development, there is a significant nagative correlation between average time interval and total number of comments and forwardings. It indicates that posting behavior is influenced by number of comments and forwardings, which unified as social concern. Based on the statistical analysis above, we presents a human dynamic model co-driven by interest and social concern, and give the analytical expression of inter-time distribution in special condition. The numerical simulation is in good accordance with analytical results.The simulation shows that the time interval between two consective messages follows a power-law distribution, which is mainly driven by the degree of users’ interests, but the interests wane as time goes by. Social concern plays a significant role regarding the change of interests, as it may slow down the decline of the latter. The higher the value of social concern, the higher frequency users post microblog messages.Secondly, the paper concerns on comment and forwarding behavior in microblog community. The Statistical result shows a significant positive correlation between number of comments and times of forwarding in loglog coordinate. Comment frequency distribution, forwarding frequency distribution and sum of comment and forwarding all follow power law distribution, the exponent of which are between the value of1and2.Compared with comment behavior, forwarding behavior in microblog community is more frequent. In addition, number of comment and forwarding has close relationship with the number of followers. The more followers the user has, the greater number of comment and forwarding the user’s message get. Based on the analysis, BA model and Poisson network model with node batch, the paper proposes a comment model including message influence and preferential attachment. In the next part, in order to verify the effectiveness of the model, we quantized parameters reasonably, and then reproduced a power-law distribution with fat-tailed feature by applying the model, which fits well with the actual data. The simulation indicates that the effect of influence mechanism is more significant than that of preferential attachment when comment attaches a message.20%of star messages get nearly70%of total comments. It reveals amazing glamour from "opinion leaders".Thirdly, the paper presents an empirical analysis and model research of the group behavior of microblog users in crisis situations. The result shows that the time interval between two consecutive messages follows a power-law distribution. But users post messages more often than they do in everyday situations and incline to post them by computer. The average length of messages in crisis situations hasn’t changed significantly; while messages related with crisis situations are posted only once by most users and always get more forwardings and comments. Furthermore, users concern a lot at the early stage of crisis situations and then their interest decays over time. As time goes by, the number of message is less and less, so does the sum of comment and forwarding. Based on the features of users’behavior in crisis situations, the paper presents comment and forwarding model for group user in crisis situations considering mechanism of preferential attachment and decay of message, comments and forwarding.The simulation result indicates that the clustering of forwarding and comments will be raised by decreasing messages and increasing the preferential attachment probability p, which make the power exponent increasing. The power exponent and the decreasing degree of forwarding and comments change in the similar trend firstly, but then in the reverse trend.Based on the conclusions above, we give suggestions on the public opinion guiding and the microblog marketing. In public opinion guiding, we emphasize the importance of monitoring time, the monitoring object and the way of guiding. In microblog marketing, we emphasize the importance of improving the message influence in microblog community, stimulating the followers’interest in posting messages and the way how to achieve better marketing effect by making good use of a register microblog ID.The innovation points in the paper are as follows:(1) Analyze user behavior in different perspectives to reveal its statistical characteristics. This data is analyzed respectively from the periodic, burst, interval time distribution, frequency distribution, correlation and many other aspects, the result of which is helpful to establish the human dynamic model;(2) Establish posting model co-driven by interest and social concern. The model is not only describing the impact of interest attenuation in posting behavior, but also further points out that the social concern is one of the key factors to change the degree of interest attenuation. Social concern and interest can stimulate user’s posting behavior;(3) Establish comment model including message influence and preferential attachment. The model quantifies followers’impact on comments attachment. It describes that message from big star and common user plays different roles in comment behavior;(4) Establish comment and forwarding model in crisis situations. The model brings in preferential attachment and quantifies message attenuation and comment attenuation. The simulation result fits well with the actual data, which can be used to explain user behavior in crisis situations. |
PDF Full Text Request