Identification method of social network spam account based on user relationship

Abstract

The invention discloses a social network garbage account identification method based on user relationship, which comprises the following steps: acquiring friend-fan relationship data of the social media account, constructing a local network relationship diagram, constructing a local symmetric network structure diagram and extracting symmetric characteristics of the network structure, and adoptinga supervised learning method to carry out garbage account identification. On the basis of constructing a user local network relation graph, as the network relationship extends from a single user network relationship to a two-way network relationship between the us and other users, the local network diagram is expanded and the new feature extraction method is adopted, which shows better stability in the garbage account identification, greatly reduces the content fraud of the garbage account identification, and makes the identification of the garbage account in the accuracy has been significantly improved.

Description

technical field [0001] The invention belongs to the technical field of data analysis, and in particular relates to a method for identifying junk accounts in social networks based on user relationships. Background technique [0002] Nowadays, in the era of advanced information, the rapid development of the Internet makes people's interaction and communication more convenient, and also shortens the communication distance between people. Social media has become an important platform and way for people to share resources and communicate with each other. People can establish their own social network system through social media, and can interact in real time with relatives, friends, classmates and colleagues in real life. Look for strangers who share personal interests, professional interests, and even shared emotions. Such a powerful function of social media makes people's social circle continue to expand, while the connection distance between each individual continues to shrink...

AI Technical Summary

Problems solved by technology

[0007] Spam accounts show increasingly strong imitation in the content of messages sent by users, and at the same time, the rules of sending messages and messages sent by users are very unstable, which has caused great difficulties in the detection of account content.

First of all, for social networks like Twitter, the length of tweets sent by users is very limited, and the content will not show too many useful features. At the same time, spam accounts gradually learn some rules of content representation of normal accounts. , so that the spam account filtering method based on content detection can be greatly deceptive by means of keyword replacement and content modification, which affects the accuracy of spam account identification; secondly, the method of establishing a local relationship network graph between users, Only considering the one-way local network relationship from one user to another user, it is one-sided in the identification of spam accounts, and cannot well interpret the connection relationship of users in the social network. Therefore, in the feature extraction of network relationships, the There are few useful features, which greatly limits the scope of use of spam account identification

Images