User Behavior Feature Mining In Film Evaluation Data

Although the quantitative study of human behavior dynamics started late,it received extensive research in recent years.Watching movies becomes one of the most popular forms of entertainment for contemporary people,especially for the young people.Studying dynamics of people's watching and evaluating film,understanding the interest taste and mining the potential pattern of users,and adjusting the recommendation for users to improve the clicks of the film networks and attract the potential customers,which would undoubtedly bring great commercial value to the film companies.However,most of the useful information is hidden in the big data.Without in-depth study and analysis,the useful information cannot be revealed.Firstly,we study the basic characteristics of film evaluation data.Then,we analyze the number of film evaluations on time,and find the film evaluations show two peaks at the 99 th and 209 th days after the first evaluation.At the 209 th day,the number of film evaluation achieves the highest number.Further,the behavior characteristics of the users at the peak and non-peak film evaluation periods are analyzed.Similar characteristics of users are found in these two stages,and their probability distributions of the user degrees obey the power-law distribution,indicating the users in both periods are mainly inactive users.Furthermore,we study the user behavior of the active and inactive evaluation users,and find that the user degree distribution of the active evaluation users obey the power-law distribution,indicating active evaluation users are mainly the small-degree users.The degree distribution of the inactive evaluation users show a relatively more uniform distribution,and the probability of users with degree ranging from 100 to 300 is higher.Finally,the behavior characteristics of the users who evaluate the popular movies and unpopular movies are studied,respectively.By analyzing the characteristics of time evolution of the film evaluation,and studying the user structure of the peak and nonpeak periods of the film,and the active and inactive evaluation users' characteristics,we find the users who evaluate the popular and unpopular films show similar behavior.