Terminal fraud phone recognition method based on call text word vector

Abstract

A terminal fraud phone recognition method based on a call text word vector comprises the steps: enabling a user to mark an incoming call in a terminal App; when the incoming call is marked as a fraudcategory, converting the incoming call into a text after user authorization agreement, and checking and desensitizing the text by the user and uploading the text to a server to be stored as a text sample after user authorization; performing word segmentation and part-of-speech tagging on the text sample to obtain syntactic dependency tags and word combination vectors of the segmented words, splicing the word combination vectors, the part-of-speech tagging and the syntactic dependency tags to form content vectors of the segmented words, and calculating scene element tags to which the segmentedwords belong to obtain semantic vectors of the text sample; constructing a fraud classification and recognition model, using a text sample in a server as a training sample, and pushing the trained model to the App from the server; and after the App receives the new to-be-identified call, obtaining a fraud-related category to which the new to-be-identified call belongs according to the model, and prompting a user. The invention belongs to the technical field of information, and can accurately identify fraudulent calls based on call texts.

Description

technical field [0001] The invention relates to a terminal fraudulent call identification method based on a call text word vector, which belongs to the field of information technology. Background technique [0002] At present, the number of telecommunications fraud cases initiated abroad is increasing day by day, and mobile phone users have higher and higher requirements for filtering fraudulent calls. However, more and more fraudsters’ call behavior tends to be concealed, and the characteristics related to call behavior are weakened. Only by analyzing and identifying call texts can it be possible to further improve the accuracy and recall rate of mobile phone systems for identifying bad calls. [0003] At present, the fraudulent call filtering method based on the mobile terminal system on the market is still relatively rudimentary. Mainstream manufacturers usually use the method of user marking, that is, relying on the user to actively mark the category of the phone and up...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that it is impossible to detect fraudulent calls in real time, and the victim is often deceived when it is discovered

Images