A method for analyzing fraudulent calls based on multidimensional time series
A time-series, fraudulent call technology, applied in neural learning methods, data processing applications, forecasting, etc., can solve problems such as threatening the security of the telecommunications network, harming the interests of telecommunications users, the reputation of the telecommunications network, and difficulty in distinguishing
Active Publication Date: 2021-08-17
NAT COMP NETWORK & INFORMATION SECURITY MANAGEMENT CENT +1
View PDF7 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0003] The current fraudulent calls are characterized by diversity, concealment, and high confrontation. They even use high-tech means to change their numbers and counterfeit the numbers of other people or organizations for fraud.
The call behavior of these fraudulent calls is very similar to that of ordinary normal calls. They are hidden in a large number of bills, and it is difficult to distinguish them. Moreover, many fraudulent calls are normal calls in the early stage, which is very confusing. Scammers have various methods of deception , Ever-changing, people are hard to guard against, this also seriously affects the call order and threatens the security of the entire telecommunication network, damages the interests of telecommunication users and the reputation of the telecommunication network
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0021] In order to enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.
[0022] The present invention is based on the CDR bill data with fraud marks, and derives multidimensional variables according to its number characteristics and behavior characteristics, and then calculates the data in the time dimension, and finally forms multidimensional time series data with labels. The multidimensional time series data is brought into the long short-term memory network (LSTM) for training, and the long-term short-term memory neural network model is established. Substituting the multi-dimensional statistical features of a number for 24 consecutive hours into the model can determine whether the number is a fraudulent call in the 25th hour.
[0023] refer to figure 1 As shown, a method for analyzing frau...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a method for analyzing fraudulent calls based on multidimensional time series. The method includes: taking all calls of each number as a whole, selecting behavioral features that are more relevant to fraudulent calls at intervals, and calculating each The characteristic statistics of the number in the interval time period, and set whether it is a label of a fraudulent call; arrange the multiple behavior characteristics of each number in the set interval time period in chronological order, and integrate them into a complete time period Multiple numbers form multiple multidimensional time series data sets within a multidimensional time series data set; Substitute multiple multidimensional time series data sets with labels into LSTM network model training; according to the model of a certain number in the complete time period Train to predict whether a call to that number is a spoofed call at the next interval of that full time period. Through the method of the invention, it is possible to analyze and predict whether it is a fraudulent call from a large number of bill data.
Description
technical field [0001] The invention belongs to the technical field of artificial intelligence and big data, and in particular relates to a method for analyzing fraudulent calls based on multidimensional time series. Background technique [0002] In recent years, there has been an explosive trend in the use of telephone fraud, with a wide range of victims and huge amounts of money. Communication fraud has become a huge pain point for users. [0003] The current fraudulent calls are characterized by diversity, concealment, and high confrontation. They even use high-tech means to change their numbers and counterfeit the numbers of other people or organizations for fraud. The call behavior of these fraudulent calls is very similar to that of ordinary normal calls. They are hidden in a large number of bills, and it is difficult to distinguish them. Moreover, many fraudulent calls are normal calls in the early stage, which is very confusing. Scammers have various methods of decep...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): H04W4/14H04M3/22G06Q10/04G06N3/08
Inventor 张震孟许歌缪亚男马欢庞韶敏李波波于芳名金红杨满智刘长永
Owner NAT COMP NETWORK & INFORMATION SECURITY MANAGEMENT CENT