Research On Statistical Analysis And Modeling Of Human Online Communication Behavior Based On QQ Group

With the continuous development of internet technology,more and more online network platforms provide great convenience for the production and life of human beings.A lot of human behavior data are recorded,which provides a good opportunity for us to more accurate statistics and analysis of human behavior rules.Many scholars have done a lot of work from the perspective of a single individual,two individuals,and got good research results,but the research on the frequent interaction of many groups is relatively less.However,human activities with frequent interactions among numerous individuals in a group are the basis of many social and economic phenomena.Our research results will provide in-depth insights into the activities of various human groups,help to understand many complex socio-economic phenomena,and produce certain application value in public opinion control,message dissemination,information recommendation,advertising,resource allocation,etc.To contribute to the development of human dynamics.In this paper,we conducted empirical statistical analysis and modeling research based on the chat data of QQ group.The specific contents of the work are as follows:(1)Empirical analysis of human behavior patterns in QQ group.Based on the chat data of QQ group,we make a statistical analysis of the online communication behavior of human beings,and conclude that the time interval between users sending consecutive messages generally presents three types: bimodal distribution,double-power-law distribution and single power-law distribution The groups with these three distribution types have significantly different levels of chat activity.From the bimodal group to the double-power-law group to the single power-law group,the level of group activity increases in turn.Meanwhile,the level of group activity affects the size of the power index,usually with as the degree of the level of group activity increases,the absolute value of the power index shows an increasing trend.Although the average daily total amount of speech of these three types of QQ groups varies greatly,the distribution form of speech time in each hour of the average day is roughly the same,presenting a periodic bimodal pattern with a period of one day.The probability distribution of chat topic length in QQ group generally presents a power-law distribution mode,which indicates that numerous topics in the group can not get a positive response,and the group’s activity degree affects the size of its power index.On the whole,with the increase of the group’s activity degree,the absolute value of the power index shows a decreasing trend,which shows that the more active the group is,the more likely it is to have long topics.The probability distribution of the number of users’ speech in each group also generally presents the characteristics of power law distribution,which reflects that most people in the group are in a relatively silent state,and only a few people are more active in chatting.The relationship between the level of group activity and the number of group members is that the level of group activity tends to increase with the increase of group size,but it is not strictly positive correlation.(2)The mathematical model of interaction behavior in QQ group is established based on empirical statistical law.Summarizing the results of empirical statistical analysis,it can be concluded that there are three main mechanisms in QQ group(i.e.task initiation rate influenced by human physiological rhythm,the level of group activity and interaction between individuals)that determine the behavior rules displayed by the group.Combined with these three basic mechanisms,we establish the interaction model of a human group,and the simulation results are consistent with the data analysis results.Using our model,it is found that at a constant initiation rate,regardless of whether the response rate is constant,the time interval distribution exhibits a bimodal distribution.These three distributions can also be obtained when the initiation rate changes under other periodic laws.The results of the study show that the periodic rhythm of human beings,together with the high degree of activity of speech,play a key role in the appearance of the double-power-law distribution mode and the single power-law distribution mode.(3)Task queue model of three individual interaction.We first improved the theoretical model of the two-person task queue.The study found that when the user in the theoretical model of the two-person task queue has at most one type I task in the task list at each moment,the model results are almost identical to the original model difference.When considering the periodic rhythm of a human being,the user’s initiation rate changes periodically and the average initiation rate increases gradually,the time interval distribution between two individuals sending messages to each other changes from bimodal distribution to double-power-law distribution and then to single power-law distribution.Once again,it shows that the periodic rhythm of human beings,together with the high degree of activity of speech,play a key role in the appearance of the double-power-law distribution mode and the single power-law distribution mode.Finally,we extend the task queue theory to the system of three individual interaction.When the user’s initiation rate changes periodically and increases with the average initiation rate,the time interval distribution between three members sending messages to each other also changes from bimodal distribution to double-power-law distribution and then to single power-law distribution.