Method and device for monitoring and identifying fraudulent behavior of FaceTime at mobile phone interface, and medium

By obtaining FaceTime data interaction records, designing abnormal call and den recognition models, combining dynamic thresholds and Stacking models, the problem of difficult FaceTime calls is solved, effectively monitoring and identification of FaceTime fraudulent behaviors is realized, and the intensity of fraud crackdowns is enhanced. .

CN119012202APending Publication Date: 2024-11-22广州市申迪计算机系统有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410975421.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The security protocol of FaceTime calls makes it difficult to be effectively supervised by external regulatory agencies and has become a hotbed of telecommunications fraud. The existing anti-fraud measures mainly rely on users' active shutdown of functions or increased vigilance, and lack of active prevention solutions.

Method used

By obtaining the data interaction records of FaceTime on the mobile phone, design an abnormal caller monitoring and identification model and den discovery model, combining dynamic monitoring thresholds and Stacking model feature indicators, it realizes monitoring and identification of FaceTime's fraudulent behavior, and outputs the number that needs to be shut down. and abnormal cell location.

Benefits of technology

The FaceTime abnormal call monitoring and fraud den discovery based on dynamic thresholds have been realized, which has improved the accuracy and comprehensiveness of identification and provided clues of fraud dens for relevant departments to crack down on.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119012202A_ABST
    Figure CN119012202A_ABST
Patent Text Reader

Abstract

The invention discloses a method and a device for monitoring and identifying fraudulent behaviors of FaceTime at a mobile phone interface, and a medium. The method comprises the following steps: acquiring a data interaction record based on a mobile phone interface FaceTime; designing prediction rules and characteristic indexes of the FaceTime abnormal calling monitoring and identification model and the FaceTime set point discovery and identification model; determining a dynamic monitoring threshold value of each identification model; performing FaceTime fraud-related behavior monitoring by using the FaceTime abnormal calling monitoring and identification model and the FaceTime fob discovery and identification model; and carrying out feature stacking on the prediction result output by each identification model to obtain a final prediction result. The mobile phone interface FaceTime abnormal calling monitoring and identification and fraud spot discovery and identification can be realized based on a dynamic threshold algorithm, the accuracy and comprehensiveness of identification are improved in combination with a Stacking model feature index fusion technology, and a number needing to be shut down is finally output; and meanwhile, an abnormal cell position can be provided as a possible fraud point clue, so that related departments can cooperatively strike FaceTime fraud-related behaviors.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of telecommunications fraud monitoring, and in particular to a method, device and medium for monitoring and identifying fraudulent behavior involving FaceTime on a mobile phone. Background Art

[0002] FaceTime is a video calling application developed by Apple that allows users of Apple devices to make video and audio calls. Since FaceTime calls use multiple security protocols such as point-to-point connection (ICE), session (SIP), and audio and video stream transmission (SRTP) to encrypt information, it is difficult for external regulatory agencies to effectively supervise FaceTime calls, making it a hotbed for telecommunications fraud.

[0003] Fraud based on FaceTime calls mainly involves strangers modifying the shared name, profile picture and chat background in advance, impersonating customer service staff of e-commerce companies, banks or social platforms, and then making one-way calls based on phone numbers or email addresses, using the personal information obtained in advance to make various excuses to get the victim to transfer money. In addition, there are also scammers who will guide users to download other conference software during FaceTime calls, obtain users' passwords, verification codes and other important information through "screen sharing", and ultimately commit fraud.

[0004] Given the security protocol protection of FaceTime, existing FaceTime anti-fraud measures mainly rely on users to actively turn off the FaceTime function or increase vigilance during FaceTime calls, but lack an implementation plan to actively prevent FaceTime fraud. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method, device and medium for monitoring and identifying fraudulent behavior involving FaceTime on a mobile phone.

[0006] The first aspect of the present invention provides a method for monitoring and identifying mobile phone FaceTime fraudulent behavior, comprising the following steps: Get data interaction records based on FaceTime on the mobile phone port; Design prediction rules and characteristic indicators for FaceTime abnormal caller monitoring and identification models and FaceTime den discovery and identification models; the characteristic indicators are at least one statistical feature of the data interaction records; Determining a dynamic monitoring threshold for each recognition model, wherein the dynamic monitoring threshold is applied to the prediction rule of the model; Use the FaceTime abnormal caller monitoring and identification model and the FaceTime den discovery and identification model to monitor FaceTime fraudulent behavior and obtain the prediction results of abnormal FaceTime call users / cells; The prediction results output by each recognition model are feature stacked to obtain the final prediction result.

[0007] Furthermore, the data interaction records specifically include FaceTime calling records, FaceTime called records, target software data interaction records, target server interaction records, target website access records, target address book interaction records, target port interaction records, and target number change records.

[0008] Furthermore, the acquisition of data interaction records based on the mobile phone FaceTime specifically includes the following steps: Obtain FaceTime call records and FaceTime call records through the operator monitoring system; Obtain target software data interaction records, target server interaction records, target website access records, and target address book interaction records through the DPI system; Obtain the target port interaction records and target number change records through the signaling monitoring system.

[0009] Furthermore, the FaceTime abnormal caller monitoring and identification model is used to identify abnormal FaceTime call users, and the FaceTime den discovery and identification model is used to identify abnormal FaceTime call cells.

[0010] Furthermore, the FaceTime abnormal caller monitoring and identification model includes an abnormal caller model, a high-risk aggregated caller model, and a high-frequency caller model; The abnormal calling model identifies abnormal FaceTime call users based on FaceTime call frequency, call recipients, and contact book interaction records; The high-risk clustered caller model identifies abnormal FaceTime call users based on the outbound call frequency, outbound call time period, outbound call object dispersion and outbound call object location of the FaceTime caller record; The high-frequency calling model identifies abnormal FaceTime call users according to the number and frequency of FaceTime calling records and FaceTime called records.

[0011] Furthermore, the FaceTime den discovery and identification model includes a den high-frequency call model and a fraud-related assistance behavior model; The high-frequency call model of the den determines abnormal FaceTime call cells based on the FaceTime call records of users in each cell, the dispersion of outbound call objects, and the target server interaction records; The fraud-related assistance behavior model determines abnormal FaceTime call cells based on the target software data interaction records, target website visit records, target port interaction records, and target number change records of users in each cell.

[0012] Furthermore, the step of determining the dynamic monitoring threshold of each recognition model specifically includes the following steps: Determine the trend term, cycle term, holiday term and error term of each characteristic indicator in the time dimension; Fill in the Prophet model parameters according to the trend item, cycle item and holiday item of each characteristic indicator; Taking the monitoring time as input, the Prophet model is used to fit the project and output the prediction interval of each characteristic indicator; The prediction rules of the FaceTime abnormal caller monitoring and recognition model and the FaceTime den discovery and recognition model are adjusted according to the prediction interval obtained by each characteristic indicator.

[0013] Furthermore, the prediction results output by each recognition model are stacked with features and implemented through a logistic regression model, which specifically includes the following steps: Performing binarization processing on the prediction result; The prediction results of the binary processing are horizontally spliced ​​to form a feature matrix; The feature matrix is ​​identified through a logistic regression model to obtain the final prediction result.

[0014] A second aspect of the present invention discloses an electronic device, comprising a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method for monitoring and identifying fraudulent behavior involving FaceTime on a mobile phone.

[0015] A third aspect of the present invention discloses a computer-readable storage medium, wherein the storage medium stores a program, and the program is executed by a processor to implement the method for monitoring and identifying fraudulent behavior involving FaceTime via a mobile phone.

[0016] The embodiment of the present invention also discloses a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device can read the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the above method.

[0017] The embodiments of the present invention have the following beneficial effects: the method, device and medium for monitoring and identifying mobile phone FaceTime fraudulent behavior can realize the monitoring and identification of abnormal mobile phone FaceTime callers and the discovery and identification of fraud dens based on the dynamic threshold algorithm, and improve the accuracy and comprehensiveness of identification by combining the Stacking model feature index fusion technology, and finally output the number that needs to be shut down. The embodiments of the present invention can also provide abnormal cell locations as possible fraud den clues for relevant departments to jointly combat FaceTime fraudulent behavior.

[0018] Additional aspects and advantages of the present invention will be given in part in the following description, and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 This is the networking schematic diagram of the mobile phone port involved in FaceTime fraud.

[0021] Figure 2 It is a schematic diagram of the basic steps of the method for monitoring and identifying fraudulent behavior involving FaceTime via a mobile phone of the present invention.

[0022] Figure 3 It is a schematic diagram of the design rules of the fraudulent FaceTime abnormal caller monitoring and identification model of the present invention.

[0023] Figure 4 It is a schematic diagram of the design rules of the FaceTime fraud den discovery and identification model of the present invention.

[0024] Figure 5 This is a diagram showing the dynamic monitoring of a user's FaceTime call times.

[0025] Figure 6 It is a schematic diagram of the structure of the electronic equipment of the present invention.

[0026] Figure 7 It is a schematic diagram of the structure of the computer-readable storage medium of the present invention. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0028] The network architecture based on FaceTime fraud in the prior art is as follows: Figure 1 As shown. The fraudsters use various encrypted communication software to get in touch with the GOIP (Gsm Over Internet Protocol, IP-based communication network data transmission) dens set up in China. Apple phone A in the GOIP den in China is used as the "card insertion terminal" and phone B is used as the "video terminal". Phone A and phone B are connected by a video cable, and the video screen of phone B can be transmitted to phone A. When implementing the fraud, the remote control terminal A in China remotely controls the phone A in China through remote control dialing software (such as commercial software such as Sunflower or self-made remote control software); remote control terminal B establishes a network video call with phone B in China; since the video screen can be transmitted between phone A and phone B in China, when using phone A to make a FaceTime call with the user, the video screen from phone B can be displayed during the call, and the video screen of phone B is shared by the remote control terminal B in China. Therefore, the fraudsters may fake various identities abroad to convey false picture information to users and defraud users.

[0029] It can be seen from the above networking that mobile phones A and B located in the country can be monitored by the monitoring system of domestic operators. Therefore, the embodiment of the present invention can analyze the fraudulent behavior of domestic accomplices based on the data interaction records of domestic operators for mobile phones and build a targeted model for countermeasures.

[0030] like Figure 2 As shown, the first embodiment of the present invention provides a method for monitoring and identifying mobile phone FaceTime fraudulent behavior, including the following steps: S1. Obtain data interaction records based on FaceTime on the mobile phone port; S2. Design prediction rules and characteristic indicators for FaceTime abnormal caller monitoring and identification model and FaceTime den discovery and identification model; the characteristic indicator is a statistical feature of at least one data interaction record; S3. Determine a dynamic monitoring threshold for each recognition model, and apply the dynamic monitoring threshold to the prediction rule of the model; S4. Use the FaceTime abnormal caller monitoring and identification model and the FaceTime den discovery and identification model to monitor FaceTime fraudulent behavior and obtain the prediction results of abnormal FaceTime call users / cells; S5. Feature stacking is performed on the prediction results output by each recognition model to obtain the final prediction result.

[0031] The following is a detailed description of the implementation process of each step: S1. Obtain data interaction records based on FaceTime on the mobile phone port.

[0032] In order to monitor the FaceTime fraudulent behavior of domestic fraud accomplices, the embodiment of the present invention needs to obtain data interaction records such as FaceTime call records, FaceTime called records, target software data interaction records, target server interaction records, target website access records, target address book interaction records, target port interaction records, and target number change records through domestic operators. Among them, the target software data interaction record refers to the interaction record of the mobile phone using various encrypted or non-encrypted communication software such as Bat, Paper Plane, QQ, and WeChat; the target server interaction record refers to the interaction record between the mobile phone and the server that carries FaceTime calls such as Apple STUN server; the target address book interaction record refers to the interaction record of the mobile phone performing read and write operations in the Apple address book; the target port interaction record refers to the interaction record of the mobile phone accessing ports such as 1069**1107 and ports such as 00447**0297, which are usually used for access to various black and gray software; the target number change record and the target website access record refer to the implementation record of sensitive operations such as the mobile phone accessing the Apple customer service website or changing the account binding number. These data interaction records can be obtained through various data monitoring systems set up by the operator platform. Among them, FaceTime caller records and FaceTime called records can be obtained through the malicious program monitoring system; target software data interaction records, target server interaction records, target website access records and target address book interaction records can be obtained through the Internet log DPI (Deep Packet Inspection) system; target port interaction records and target number change records can be obtained through the signaling monitoring system. After obtaining the corresponding data interaction records, the embodiment of the present invention can use the corresponding recognition model to identify the mobile phone that generates the data and analyze the mobile phone involved in fraudulent behavior.

[0033] S2. Design prediction rules and characteristic indicators for the FaceTime abnormal caller monitoring and recognition model and the FaceTime den discovery and recognition model. In the embodiment of the present invention, the FaceTime abnormal caller monitoring and recognition model is used to identify abnormal FaceTime call users, and the FaceTime den discovery and recognition model is used to identify abnormal FaceTime call cells. Different recognition models have different characteristic indicators, which may be some data interaction records themselves, or may be some statistical features such as frequency, period, number of times, etc. obtained by statistical analysis based on data interaction records.

[0034] like Figure 3 As shown, the FaceTime abnormal caller monitoring and identification model of the embodiment of the present invention includes an abnormal caller model, a high-risk aggregated caller model and a high-frequency caller model.

[0035] Abnormal call model: The abnormal call model identifies possible abnormal FaceTime calls by analyzing the user's FaceTime call frequency, call recipients, contact management, and other abnormal behaviors. The specific rules are as follows: a. Frequent and high-discretion FaceTime users within the network (specific indicators include the number of calls, called dispersion, etc.) b. Suspicious users who store a large number of Apple contacts in a short period of time; (Specific indicators include the number of stored Apple contacts, etc.) c. Accompanying QQ voice / video behavior. (Specific indicators include the number of QQ voice calls, the number of QQ video calls, etc.) High-risk clustered caller model: The high-risk clustered caller model identifies possible high-risk clustered FaceTime calls by analyzing the user's FaceTime outbound calling behavior, the discreteness of the outbound calling object, the C (base station cell) or P clustering, and the abnormal conditions of the outbound calling time period. The specific rules are as follows: a. Discrete outbound calls to the same number in the same period; (Specific indicators include the number of discrete outbound calls to the same number in the same period, etc.) b. Geographically clustered outbound calls: a large number of calls under the same base station location information CI, or a large number of calls under the same overseas IP in the outbound call results are detected; (Specific indicators include the number of calls under the same base station, the number of calls under the same overseas IP, etc.) c. Abnormal outbound call time period (FaceTime behavior occurs in the early morning or midnight). (Specific indicators include the number of abnormal outbound call time periods, etc.) High-frequency calling model: The high-frequency calling model analyzes the number of FaceTime calls and calls made by the user, as well as the high-frequency calling behavior in a short period of time, combined with the abnormal ratio of callers to callers, to identify possible high-frequency abnormal FaceTime calls. The specific rules are as follows: a. The number of FaceTime calling and receiving calls on the same day is higher than the threshold number, for example, more than X; (Specific indicators include the number of FaceTime calling and receiving calls on the same day, etc.) b. The FaceTime call threshold within a single hour is high, for example, it exceeds Y; (Specific indicators include the number of FaceTime call thresholds within a single hour, etc.) c. The ratio of FaceTime callers to callers on that day is abnormal. (Specific indicators include the ratio of FaceTime callers to callers on that day, etc.) like Figure 4As shown, the FaceTime den discovery and identification model of the embodiment of the present invention includes a den high-frequency call model and a fraud-related assistance behavior model. Unlike the FaceTime abnormal caller monitoring and identification model, which analyzes the data interaction records of each mobile phone, the FaceTime den discovery and identification model performs feature clustering analysis based on all data interaction records in a cell as a whole. When multiple numbers in the same cell show fraud-related features on the same day, the cell is marked as an abnormal cell and may be a fraud-related FacTime den.

[0036] High-frequency call model of dens: The high-frequency call model of dens identifies and marks possible fraud dens by analyzing the abnormal increase in the frequency and volume of users' FaceTime calls, the discreteness of the peer numbers, and the interaction ratio with Apple's STUN server. The specific rules are as follows: a. The number of calls on that day is high, for example, more than X times; (Specific indicators include the number of calls on that day, etc.) b. The call volume on that day is high, for example, more than M times the average call volume in the past N days; (specific indicators are the same as above) c. The calling party's discreteness is high; (Specific indicators include calling party's discreteness, etc.) d. The number of interactions with Apple's STUN server on that day accounted for more than Y% (Y = 100*STUN server interaction times / (interaction times with Apple STUN server + interaction times with the peer number’s public network address + interaction times with the peer number’s private network address)). (Specific indicators include the percentage of interactions with Apple’s STUN server on that day, etc.) Fraud-related assistance behavior model: The fraud-related assistance behavior model analyzes users' FaceTime usage behavior, use of black and gray software, customer service phone call records, Apple customer service website visit records, and terminal number change behavior, and performs cluster analysis on communities to identify suspected fraud-related assistance dens. The specific rules are as follows: a. After FaceTime is registered and activated, there are interaction records with ports 1069**1107 and 00447**0297 in the signaling or call logs; (Specific indicators include the number of interactions with ports 1069**1107 and 00447**0297, etc.) b. Have a record of using other black and gray software such as Bat, Paper Plane, etc. (Specific indicators include the number of times other black and gray software has been used, etc.) c. Have a record of calling Apple customer service to request a reconnection (customer service phone number 4006668800); (Specific indicators include the number of times Apple customer service was called, etc.) d. Visited Apple's online customer service website (https: / / getsupport.apple.com / consent); (Specific indicators include the number of visits to Apple's online customer service website, etc.) e. The terminal where the above services are performed has a number change record in the past X days, and the number change frequency is greater than Y. (Specific indicators include the number of number changes, etc.) Summarizing the characteristic indicators of the recognition model in step S2, the following Table 1 can be obtained: Table 1

[0037] S3. Determine a dynamic monitoring threshold for each recognition model, and apply the dynamic monitoring threshold to the prediction rule of the model.

[0038] After determining the model rules for monitoring and identifying abnormal FaceTime callers involved in fraud and the model rules for discovering and identifying FaceTime dens involved in fraud, we need to further determine the model's monitoring threshold. The traditional fixed-threshold model detection method has shortcomings such as low accuracy and inability to adapt to complex and changeable fraud behaviors. It is difficult to adapt to today's constantly updated fraud methods and monitoring scenarios of fraud-related behaviors under massive data.

[0039] To this end, the embodiment of the present invention adopts a time series prediction algorithm based on FBProphet, which can learn historical monitoring thresholds so that the monitoring thresholds set in the model rules fluctuate according to different periods such as weekdays, weekends, special holidays, etc., and predict the future upper and lower limits of the thresholds of each characteristic indicator. By comparing the predicted interval with the actual value, abnormal behavior can be discovered.

[0040] Specifically, the Prophet algorithm used in the present invention adopts the form of a decomposable model to decompose the time series into four main components. The mathematical principle of the algorithm model is as follows: ; The model as a whole consists of four parts: g(t) represents the trend term growth, which represents the trend of change of the time series in the non-periodic state; it reflects the overall direction of the time series, whether it is rising or falling over time.

[0041] s(t) represents the periodic term, or seasonal term, which is generally measured in weeks or years. It captures the periodic patterns in the data, such as weekly or monthly cyclic trends.

[0042] h(t) represents the holiday term holidays (the impact of holidays on the predicted value), which is simulated using Fourier series to represent the impact of potential holidays with non-fixed periods in the time series on the predicted value; it captures the characteristic trend of non-fixed period holidays in the data.

[0043] That is, the error term or residual term represents the fluctuations not predicted by the model and follows a Gaussian distribution.

[0044] The Prophet algorithm of the embodiment of the present invention can obtain the predicted value of the time series by fitting the trend term, the cycle term, the holiday term and the error term and finally accumulating them.

[0045] The trend items are divided into linear growth and nonlinear growth: (1) Linear growth: based on piecewise linear function.

[0046] ; k represents the growth rate, represents the change in growth rate, and m represents the offset.

[0047] (2) Nonlinear growth: based on logistic regression function.

[0048] ; , , ; , , ; , , ; Represents carrying capacity: It is a function that changes over time and limits the maximum value that can grow; k represents the growth rate.

[0049] Holiday Items: ; in and ; The above decomposition method enables FBProphet to deeply understand the data structure and generate accurate feature indicator threshold predictions.

[0050] For example, Figure 5This is a diagram of the dynamic monitoring of a domestic accomplice FaceTime fraud number. This number is a fraud card for an old user. The original user is an Apple user. The number of FaceTime uses is not high every day, and the number of uses on weekdays, weekends, and special holidays is different. On May 26, 2024, due to the temptation of high returns, the mobile phone card was sold to the fraud gang. On May 27, 2024, the fraud gang mailed the collected number cards to the next fraud point for centralized card testing (a FaceTime call was made on May 28, 2024), and then mailed them to the next fraud den to commit fraud (multiple FaceTime calls were made on May 29, 2024). In the fraud detection of this number, if a fixed threshold model is used, there is often a one-size-fits-all situation, which is difficult to adapt to the characteristics of different model thresholds in different periods such as weekdays, weekends, and special holidays. The dynamic threshold algorithm can learn historical behavior characteristics, predict the future upper and lower limits of each indicator, and detect abnormal behavior by comparing the predicted interval with the actual value, timely and accurately discover fraudulent numbers, and avoid the drawbacks of fixed thresholds. (For example, on May 27 and 28, 2024, the number of FaceTime calls detected was lower than the lower limit of the dynamic threshold, and on May 29, 2024, the number of FaceTime calls detected was higher than the upper limit of the dynamic threshold) S4. Use the FaceTime abnormal caller monitoring and identification model and the FaceTime den discovery and identification model to monitor FaceTime fraudulent behavior and obtain the prediction results of abnormal FaceTime call users / cells; The recognition model of the embodiment of the present invention performs periodic predictions on the characteristic indicators involved in Table 1 (e.g., predictions every 1 hour). Each time, the model collects data interaction records for a period of time before the current time (e.g., 365 days) for prediction, and obtains the characteristic indicator prediction interval for a period of time in the future (e.g., 1.5 hours). After obtaining the prediction interval, when the end time of the characteristic indicator prediction interval is reached, the deviation between the predicted value and the true value of the characteristic indicator in the prediction interval is compared. If the deviation exceeds the dynamic monitoring threshold of the indicator, the characteristic indicator is judged to be in an abnormal state.

[0051] Taking the FaceTime call count as an example, the model output fields are shown in Table 2: Table 2

[0052] Among them, the two fields of true value and abnormality are real-time backfill information. Taking the output of this indicator as an example, it can be found that the dynamic threshold model can effectively identify typical domestic FaceTime fraud. The number of FaceTime calls detected on May 27, 2024 (mailing stage) and 28 (card trial stage) was lower than the lower limit of the dynamic threshold, and the number of FaceTime calls detected on May 29, 2024 (involved stage) was higher than the upper limit of the dynamic threshold.

[0053] For the FaceTime abnormal call monitoring and identification model, if all the characteristic indicators of a mobile phone are in an abnormal state, the mobile phone is judged to be an abnormal FaceTime call user and may be used by fraudsters for FaceTime fraud. For the FaceTime den discovery and identification model, if the proportion of characteristic indicators in an abnormal state in a community is higher than a certain threshold, it means that there are a large number of abnormal FaceTime call users in the community, and there may be FaceTime fraud dens hidden in the community.

[0054] S5. Feature stacking is performed on the prediction results output by each recognition model to obtain the final prediction result.

[0055] In actual operation, the embodiment of the present invention has at least 20 model feature indicators that need to be monitored. In order to improve the comprehensiveness and accuracy of monitoring, the embodiment of the present invention comprehensively stacks the 20 model feature indicators to comprehensively and effectively monitor the FaceTime numbers and dens involved in fraud, and achieve accurate crackdown on FaceTime fraud.

[0056] Step S5 specifically includes the following steps: S5-1. Binarize the prediction results. The embodiment of the present invention uses multiple pre-trained model feature indicators to predict the validation set and the training set to obtain a preliminary prediction probability. The prediction probability is binarized here (for example, the indicator is abnormal and the value is 1, otherwise it is 0). In this step, the threshold can be adjusted according to actual needs, with the purpose of simplifying the input feature space of the subsequent meta-model.

[0057] S5-2. The prediction results of the binary processing are horizontally spliced ​​to form a feature matrix. The embodiment of the present invention uses the prediction results (whether abnormal or not) of multiple model feature indicators as new features. For the validation set and the training set, they are horizontally spliced ​​through the feature stacking model to form a new feature matrix. This process actually creates a "knowledge fusion" at the feature level, so that the meta-model can learn complementary information between different models, which can significantly improve the accuracy of the prediction.

[0058] S5-3. The feature matrix is ​​identified by a logistic regression model to obtain the final prediction result. The embodiment of the present invention uses a logistic regression model as a meta-model, which is trained on a new feature set consisting of the prediction results of the basic models, with the purpose of learning how to integrate the outputs of these basic models to make a final decision. Subsequently, the same meta-model is applied to the new feature set of the test set to generate the final prediction. In the embodiment of the present invention, other models such as neural networks and support vector machines can also be used as meta-models to explore the meta-model structure that best suits the current data set and task.

[0059] For example, the evaluation results of the feature indicators before and after fusion are shown in Table 3: Table 3 Serial number Evaluation Metrics Accuracy Accuracy False Injury Rate Recall 1 Fusion of 20 feature indicators 99.57% 99.04% 0.96% 92.19% 2 FaceTime calls 97.82% 91.86% 8.14% 83.00% 3 FaceTime call dispersion 88.18% 84.52% 5.48% 74.53% 4 Number of Apple Contacts numbers stored 87.98% 82.03% 4.97% 74.84% 5 QQ Voice Number 88.18% 83.48% 6.52% 75.60% 6 QQ video times 88.18% 82.62% 7.38% 76.52% 7 Number of discrete outbound calls to the same number in the same period 88.16% 82.47% 7.53% 76.52% 8 Number of FaceTime calls on the same base station 89.16% 81.47% 8.53% 74.52% 9 Number of calls under the same overseas IP 88.11% 81.42% 7.51% 76.98% 10 Abnormal number of outbound calls during the time period 88.16% 81.00% 8.00% 78.21% 11 The threshold number of FaceTime calls and calls received on the same day 88.09% 80.16% 7.84% 78.36% 12 FaceTime call threshold within one hour 88.20% 81.17% 8.83% 78.51% 13 The ratio of FaceTime callers to callers on the same day 88.04% 79.35% 9.65% 88.67% 14 FaceTime calls on the day 88.00% 78.47% 11.53% 89.28% 15 FaceTime caller dispersion 77.87% 76.94% 13.06% 89.74% 16 Percentage of interactions with Apple's STUN server on that day 58.09% 20.16% 9.84% 88.36% 17 Number of interactions with ports 1069**1107 and 00447**0297 48.20% 61.17% 8.83% 88.51% 18 Number of times other black and gray software was used 58.04% 59.35% 10.65% 58.67% 19 Number of calls to Apple customer service 38.00% 48.47% 11.53% 69.28% 20 Number of visits to Apple customer service online website 67.87% 66.94% 13.06% 79.74% 21 Number of number changes 78.09% 50.16% 9.84% 78.36% As can be seen from Table 3, by fusing the prediction results, the fusion results have better accuracy, precision, false positive rate and recall rate than the dynamic threshold monitoring of a single feature indicator. Especially when dealing with complex data features and patterns, multi-model fusion shows better performance stability and prediction ability. Therefore, in practical applications, a reasonable model feature indicator fusion dynamic threshold monitoring strategy is an effective way to improve the efficiency and accuracy of FaceTime fraud number recognition.

[0060] Figure 6 It is a schematic diagram of the structure of the electronic device proposed in the second embodiment of the present invention. The memory of this embodiment stores program instructions for implementing the method for monitoring and identifying fraudulent behavior via FaceTime via mobile phone port in any of the above embodiments. The processor is used to execute the program instructions stored in the memory to monitor and identify fraudulent behavior via FaceTime via mobile phone port. Among them, the processor can also be called CPU (Central Processing Unit). The processor may be an integrated circuit chip with signal processing capabilities. The processor can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0061] The contents of the method in the first embodiment of the present invention are all applicable to the electronic device embodiment. The functions specifically implemented by the electronic device embodiment are the same as those of the above method embodiment, and the beneficial effects achieved are also the same as those achieved by the above method.

[0062] Figure 7This is a schematic diagram of the structure of the computer-readable storage medium of the third embodiment of the present invention. The computer-readable storage medium of the fourth embodiment of the present invention stores program instructions that can realize the monitoring and identification of the above-mentioned mobile phone FaceTime fraudulent behavior, wherein the program instructions can be stored in the above-mentioned storage medium in the form of a software product, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of the various implementation methods of the present invention. The aforementioned computer-readable storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes, or terminal devices such as computers, servers, mobile phones, tablets, etc.

[0063] The contents of the method in the first embodiment of the present invention are all applicable to the computer-readable storage medium embodiment. The functions specifically implemented by the computer-readable storage medium embodiment are the same as those of the above method embodiment, and the beneficial effects achieved are also the same as those achieved by the above method.

[0064] This embodiment also provides a computer program product. When the computer program product runs on a computer, it enables the computer to execute the above-mentioned related steps to achieve the monitoring and identification of fraudulent behavior via FaceTime via the mobile phone provided in the above embodiment.

[0065] Those skilled in the art will appreciate that the modules in the devices in the embodiments of the present invention may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments of the present invention may be combined into one module or unit or component, and in addition they may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the corresponding claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the corresponding claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0066] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0067] It should be noted that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0068] In addition, each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. In particular, for embodiments such as devices and equipment, since they are basically similar to method embodiments, the relevant parts can refer to the partial description of the method embodiments. The embodiments of the devices and equipment described above are merely schematic, wherein the modules, units, etc. described as separate components may or may not be physically separated, that is, they may be located in one place, or they may be distributed in multiple places, such as nodes of a system network. Specifically, some or all of the modules and units may be selected according to actual needs to achieve the purpose of the above-mentioned embodiment scheme. Those skilled in the art can understand and implement it without paying creative labor.

[0069] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0070] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0071] In addition, the terms "first", "second", etc. used in the embodiments of the present invention are only used for descriptive purposes and should not be understood as indicating or implying relative importance, or implicitly indicating the number of technical features indicated in the present embodiment. Therefore, the features defined by the terms "first", "second", etc. in the embodiments of the present invention can explicitly or implicitly indicate that the embodiment includes at least one of the features. In the description of the present invention, the word "multiple" means at least two or two or more, such as two, three, four, etc., unless otherwise clearly and specifically defined in the embodiments.

[0072] In the embodiments of the present invention, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element. In addition, components, features, and elements with the same name in different embodiments of the present invention may have the same meaning or different meanings, and their specific meanings need to be determined by their explanation in the specific embodiment or further combined with the context in the specific embodiment.

[0073] Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and cannot be construed as limitations of the present invention, and those of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present invention. Those skilled in the art will readily come to think of other embodiments of the present invention after considering the specification and practicing the present invention. This application is intended to cover any variations, uses or adaptations of the present invention, which follow the general principles of the present invention and include common knowledge or customary technical means in the art that are not disclosed in the present invention. The specification and embodiments are intended to be exemplary only, and the true scope and spirit of the present invention are indicated by the following claims.

Claims

1. A method for monitoring and identifying fraudulent behavior involving FaceTime on a mobile phone, characterized in that: The following steps are involved: Get data interaction records based on FaceTime on the mobile phone port; Design prediction rules and characteristic indicators for FaceTime abnormal caller monitoring and identification models and FaceTime den discovery and identification models; The characteristic indicator is at least one statistical characteristic of the data interaction record; Determining a dynamic monitoring threshold for each recognition model, wherein the dynamic monitoring threshold is applied to the prediction rule of the model; Use the FaceTime abnormal caller monitoring and identification model and the FaceTime den discovery and identification model to monitor FaceTime fraudulent behavior and obtain the prediction results of abnormal FaceTime call users / cells; The prediction results output by each recognition model are feature stacked to obtain the final prediction result.

2. The method for monitoring and identifying mobile phone FaceTime fraudulent behavior according to claim 1, characterized in that: The data interaction records specifically include FaceTime calling records, FaceTime called records, target software data interaction records, target server interaction records, target website access records, target address book interaction records, target port interaction records, and target number change records.

3. The method for monitoring and identifying mobile phone FaceTime fraudulent behavior according to claim 2, characterized in that: The acquisition of data interaction records based on the mobile phone FaceTime specifically includes the following steps Obtain FaceTime call records and FaceTime call records through the operator monitoring system; Obtain target software data interaction records, target server interaction records, target website access records, and target address book interaction records through the DPI system; Obtain the target port interaction records and target number change records through the signaling monitoring system.

4. The method for monitoring and identifying mobile phone FaceTime fraudulent behavior according to claim 1, characterized in that: The FaceTime abnormal caller monitoring and identification model is used to identify abnormal FaceTime call users, and the FaceTime den discovery and identification model is used to identify abnormal FaceTime call cells.

5. The method for monitoring and identifying mobile phone FaceTime fraudulent behavior according to claim 4, characterized in that: The FaceTime abnormal caller monitoring and identification model includes an abnormal caller model, a high-risk aggregated caller model, and a high-frequency caller model; The abnormal calling model identifies abnormal FaceTime call users based on FaceTime call frequency, call recipients, and contact book interaction records; The high-risk clustered caller model identifies abnormal FaceTime call users based on the outbound call frequency, outbound call time period, outbound call object dispersion and outbound call object location of the FaceTime caller record; The high-frequency calling model identifies abnormal FaceTime call users according to the number and frequency of FaceTime calling records and FaceTime called records.

6. The method for monitoring and identifying mobile phone FaceTime fraudulent behavior according to claim 4, characterized in that: The FaceTime den discovery and identification model includes a den high-frequency call model and a fraud-related assistance behavior model; The high-frequency call model of the den determines abnormal FaceTime call cells based on the FaceTime call records of users in each cell, the dispersion of outbound call objects, and the target server interaction records; The fraud-related assistance behavior model determines abnormal FaceTime call cells based on the target software data interaction records, target website visit records, target port interaction records, and target number change records of users in each cell.

7. The method for monitoring and identifying mobile phone FaceTime fraudulent behavior according to claim 1, characterized in that: Determining the dynamic monitoring threshold of each recognition model specifically includes the following steps: Determine the trend term, cycle term, holiday term and error term of each characteristic indicator in the time dimension; Fill in the Prophet model parameters according to the trend item, cycle item and holiday item of each characteristic indicator; Taking the monitoring time as input, the Prophet model is used to fit the project and output the prediction interval of each characteristic indicator; The prediction rules of the FaceTime abnormal caller monitoring and recognition model and the FaceTime den discovery and recognition model are adjusted according to the prediction interval obtained by each characteristic indicator.

8. The method for monitoring and identifying mobile phone FaceTime fraudulent behavior according to claim 1, characterized in that: The feature stacking of the prediction results output by each recognition model is implemented by a logistic regression model, which specifically includes the following steps: Performing binarization processing on the prediction result; The prediction results of the binary processing are horizontally spliced ​​to form a feature matrix; The feature matrix is ​​identified through a logistic regression model to obtain the final prediction result.

9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store programs; The processor executes the program to implement a method for monitoring and identifying fraudulent behavior involving FaceTime via a mobile phone as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement a method for monitoring and identifying fraudulent behavior involving FaceTime via a mobile phone as described in any one of claims 1-8.