Social network user friend influence measurement method based on tool variable method

Abstract

A social network user friend influence measurement method based on a tool variable method includes the following steps: S1: extracting friend information, attribute features and behavior characteristics of users in a social network, and selecting appropriate tool variables according to specific problems; S2: classifying the user friends according to the specific problem; S3: using the tool variables to perform first-stage least squares regression, and estimating the activity amount of different types of user friends; S4: using the estimated value of the user friend activity obtained in S3 to perform second-stage least squares regression on the users' own activity amount, and taking an obtained regression coefficient as a measure of the influence size. According to the social network user friend influence measurement method, the tool variable method is utilized to decouple internal factors and external factors that affect the user's activities, an unbiased estimation of the user's activity amount is made, so that a more accurate quantitative description of the user influence in the social network is achieved, and the social network user friend influence measurement method has a practical application value.

Description

technical field [0001] The invention relates to the technical fields of data mining and data analysis, in particular to a method for measuring the influence of friends in a social network based on an instrumental variable method. Background technique [0002] The rapid development of social economy and Internet technology has promoted the vigorous development of online social networks, and at the same time, the massive user data generated by it has become an important resource for data mining research and applications. Different behaviors of users on social networks often have an impact on other users in the network, and users with different attributes have different influences. Therefore, how to explore the influence mode between users and find influential users in the massive user data is of great significance for the study of information dissemination methods and the precise placement of commercial advertisements, and also has a wide range of social Value and rich commer...

AI Technical Summary

Problems solved by technology

[0003] The main problem in the current research on the mutual influence between users in social networks is that a variety of external factors and internal factors will have an impact on user behavior, which makes it difficult to determine the relative influence between users.

Images