Prediction Method of Communication Network User Calling Object Based on Representation Learning and Behavior Features

Abstract

The invention provides a communication network user call object prediction method based on representation learning and behavior characteristics, which belongs to the technical field of communication network user behavior analysis, including: randomly generating the initial vector of each communication node in the communication network, constructing a call sequence network; Extract triplet sets from historical records, construct and train a graph representation learning model stacked by LSTM cyclic neural network and bilinear layer, and obtain updated communication node vectors of all communication nodes; construct and train by parallel communication duration‑ Position weighting layer and bidirectional long-term short-term memory neural network, stacking feed-forward neural network and bilinear layer call object prediction model; the historical call records of users to be predicted are based on the updated communication node vector and the call object prediction model after training, Fulfill predictions. The invention is based on the analysis of user call history records, and realizes prediction according to communication node topology information, communication sequence information and long-term and short-term behavior characteristics of users.

Description

Technical field[0001]The present invention belongs to the field of communication network user behavior analysis, and specifically, the present invention relates to a communication network user call object prediction method based on representation and behavioral characteristics.Background technique[0002]With the update of information technology, the mobile communication tools in recent years have become rapidly popular, and greatly facilitating people's lives and producing massive user communication behavior history data. However, the convenient communication tool has also become the linkage and organization of the criminal gang. Due to the hiddenness of the criminal gang organizes and the lag of the law enforcement personnel, the prediction of the next communication target in communication network has become a very important and Valuable questions, can effectively assist relevant law enforcement departments to accurately predict the future possible communication objects of target us...

AI Technical Summary

Problems solved by technology

[0003] However, there are many difficulties in predicting the target user's next communication object in the large-scale user call historical behavior data: First, the diversified services of telecom operators and the rapid growth of user data have produced massive user data, often a market There are tens of millions of users in the communication data within the level range, and the large amount of calculation makes it difficult to effectively deal with traditional methods; in addition, users are not independent individuals in the communication network, but are associated and form a communication network, and The user's call history contains information such as communication priority, and the communication object prediction method that only considers the statistical characteristics of the user's call history record or the communication object prediction method that only considers the user's node similarity in the communication network has limited capabilities; in addition, the existing The research on user behavior in the communication network mainly focuses on the analysis of the overall traffic change of the communication network, the identification of counterfeit numbers, and the detection of fraudulent calls. There is a lack of ready-made technical solutions and in-depth research on the prediction of the next communication object of the user

Images