Potential semantic index-based internet surfing behavior analysis method and system, and medium

Abstract

The invention provides a potential semantic index-based internet surfing behavior analysis method and system, and a medium. The method comprises the following steps: step S1, determining a potential semantic feature matrix of an Internet surfing behavior of a user based on a historical Internet surfing log of the user; s2, calculating a behavior chain of the Internet surfing behavior of the user by utilizing the potential semantic feature matrix; and S3, constructing an LSTM deep neural network model according to the behavior chain so as to detect the abnormal Internet surfing behavior of the user. According to the method, the internet log data can be effectively analyzed, correlation calculation is carried out on features of websites, internet behavior portraits of users are described, an internet abnormal behavior chain is constructed, potential features are deeply mined by adopting machine learning, abnormal behaviors in the data are identified, continuous iteration, optimization and continuous improvement can be carried out, and therefore, timely emergency response and disposal are realized.

Description

technical field [0001] The invention relates to the field of semantic indexing, in particular to a method, system and medium for analyzing online behavior based on latent semantic indexing. Background technique [0002] The Internet brings great convenience to work. Without effective management, it will bring various risks to the business, such as information leakage, network attacks, and low work efficiency. The analysis of user online behavior based on the semantic features of webpages takes the user as the perspective, and changes from rule analysis to association analysis, behavior modeling, and anomaly analysis. Correlate and analyze abnormal behaviors such as internal violations, data theft, and illegal deletions with normal behaviors, model behaviors, and accurately describe behavior details, thereby improving the accuracy of hitting abnormal events and making up for the inability of traditional behavior analysis to monitor internal Insufficient threats, effective an...

AI Technical Summary

Problems solved by technology

The above method of using URL categories to analyze online behaviors relies on category determination rules, which are easily disturbed by category noise, and the definition of labels cannot fully explain the abnormal behavior of users.

Images