Machine learning method, apparatus, system, storage medium and equipment

Abstract

The present invention discloses a machine learning method, a machine learning device, a system, a storage medium and an apparatus. The machine learning method, the machine learning device, the system,the storage medium and the apparatus are used for identifying malicious network bots. The method includes an obtaining step and a learning step. According to the obtaining step, user access log datain a target website are obtained, wherein the user access log data contain the network addresses of user requests and time intervals between the requests. According to the learning step, machine learning is performed on the user access log data to obtain a time interval model of the requests. According to the method provided by the technical schemes of the invention, an invasion technology is notneeded; network abnormal bots can be effectively identified; and the accuracy of the detection of the abnormal bots can be remarkably improved.

Description

technical field [0001] The invention relates to the field of data processing of computer networks, in particular to a machine learning method, device, system, storage medium and equipment. Background technique [0002] At present, more than 50% of website traffic comes from web bots (Robots), and only about one-third of these web traffic is normal traffic, such as crawlers of search engines. More abnormal traffic is used for abnormal requests such as malicious crawlers, cc attacks, using business vulnerabilities to swipe interfaces, scans, etc. This also directly leads to the detection of abnormal bots, which is an urgent problem in the industry. [0003] To solve this problem, most of the solutions currently adopted in the industry are active detection methods, using intrusive SDK (Software Development Kit, software development kit) to collect specific information and then analyze and judge. [0004] The bot itself is a computer program, and the intrusive SDK method is us...

AI Technical Summary

Problems solved by technology

This technology also has three problems: first, the bot can return fingerprints through browser plug-ins or browser modifications; second, the device fingerprint is returned by the bot to the server, so the bot can modify

The current communication system does not popularize server-to-client identity verification, so it is impossible to introduce asymmetric encryption technology to ensure

Third, fingerprint conflicts. With the unified processing of cloud computing and IT infrastructure, the hardware configuration, operating system version and application software version of the machine are completely consistent, resulting in an increase in the probability of fingerprint conflicts and a decrease in accuracy.

For example, a bot randomly initiates a simulated human request for certain patterns of resources, and it is difficult for behavior models to effectively identify them.

[0009] On the other hand, the intrusive SDK solution is expensive to deploy, and needs to be modified to access the SDK and modify the original business code

Affect the security and privacy of the original business

Images