Personal harmful call detection method and device based on streaming data graph and readable medium

Abstract

The invention provides a personal harmful call detection method and device based on a streaming data graph and a readable medium, and the method comprises the steps: collecting CDR data of a pluralityof users, decomposing the CDR data to obtain communication data, storing the communication data in a graph database, and obtaining a communication graph; communication node statistics and related attributes and representations of the sub-graphs are extracted from the communication graph based on the depth graph; preprocessing the related attributes and the representations, training the related attributes and the representations as machine learning model parameters to obtain a first analysis result, and classifying suspicious call groups in the first analysis result; storing the suspicious call group into a graphic database, and carrying out secondary analysis to obtain the number of the harmful call. The method is advantageous in that real-time performance and accuracy are improved to a large extent, current and historical graph characteristics and record traceability of the call can be checked in real time, manual confirmation is carried out in the mode of repeatedly listening to thecall, and the analysis success rate is increased.

Description

technical field [0001] The invention belongs to network science and telecommunication network data analysis, and in particular relates to a personal harmful call detection method, device and readable medium based on flow data graph. Background technique [0002] With the popularity of wireless communication technology, mobile phones have become an essential way for people to make long-distance calls, bringing great convenience to life. However, everything has two sides. While wireless communication technology provides convenience for life, it is also easy to be used by criminals to commit crimes. Telecom fraud is a new type of fraud derived from the popularization of mobile communications. Fraud methods have caused huge losses to life and economy, and how to effectively detect harmful calls is imminent. [0003] At present, the widely used harmful call detection technology mostly discovers the potential trend of telecom fraud by analyzing the calling user's behavior. This...

AI Technical Summary

Problems solved by technology

However, the above-mentioned methods all rely on post-processing analysis. Although they have a certain effect on the prevention of telecom fraud, they cannot achieve the expected effect.

In addition, the existing methods are not very accurate in detecting fraudulent gangs, and cannot timely and accurately analyze the location of fraudulent gangs and their members from massive data

Images