A method of associating accounts in an online social network

Abstract

The invention discloses a method for associating accounts in multiple online social networks (OSNs) based on node similarity, which extracts friend relationships from OSN website accounts to associate multiple accounts belonging to the same entity user on different OSN platforms. The patent has five parts: the data preprocessing part preprocesses the account node relationship graph of OSN; the node sequence extraction part obtains the account node sequence set through random walk; the account vector representation part generates each account through the word vector tool word2vec Vector model; calculate the linear transformation matrix part to obtain the linear transformation matrix W from one OSN to another OSN through the gradient descent method; obtain the associated account part to map the account in one OSN to the coordinate space of another OSN, through the similarity Metrics and threshold screening to obtain associated accounts or corresponding candidate sets corresponding to all accounts. The invention avoids the error caused by the account association result caused by the unauthentic feature information of the account, thus improving the robustness of the account association.

Description

technical field [0001] The invention belongs to the technical field of network information, and more specifically relates to a method for associating accounts in an online social network. Background technique [0002] Account association is the main technical method used to mine all possible account information of a user in various online social networks. It can be used to track malicious users who commit cybercrime to prevent the spread of malicious events, deeply mine account information to optimize recommendation methods, and detect After the account information is stolen, a danger warning will be issued to protect the account security of other websites. [0003] The features extracted by traditional account association methods include the personal information of the account itself such as gender, age, and geographic location, as well as user behavior rules such as text writing style, search term usage habits, and mouse dwell time. However, due to personal privacy, perso...

AI Technical Summary

Problems solved by technology

Although the traditional account association method completes the task of account association, the traditional account association method requires a lot of data input information, and it is difficult to avoid the deviation of the association result caused by false account information, so the robustness of the account association is not high.

Images