Anti-fraud call method, device and equipment

Abstract

The invention provides an anti-fraud call method, device and equipment. The method comprises the following steps: a core service module receives a first request message sent by an AS module, wherein the first request message is used for requesting response information when the AS module simulates a virtual holder of a first number to communicate with a first terminal; executing a first operation according to the call context information and the configuration information of the virtual holder; wherein the first operation comprises at least one of the following items: generating response information; fraud detection is carried out through semantic analysis; carrying out data collection; and carrying out number labeling on the first terminal. According to the method, the suspected fraud code number can be actively discriminated and found in advance, the fraud risk is processed in a pre-judgment mode and measures are taken on the premise of low cost and no actual loss, and the fraud recognition accuracy is improved.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to an anti-fraud call method, device and equipment. Background Art

[0002] Fraudsters contact victims through online media, phone calls, text messages, social tools and other channels and use a variety of means to commit fraud. Currently, the common anti-fraud methods in telecommunications networks are:

[0003] By comparing the incoming call number labels with those in the big data application software on the client side, it is determined whether the current incoming call number is suspected of fraud, and the call is blocked before it is connected. This method has been found to have mislabeling or deliberate mislabeling for commercial purposes, and there is a risk of intentional or unintentional misjudgment.

[0004] By extracting the call content of the communication business, the traditional large-scale model intention recognition is used to determine whether the call content involves fraudulent content and to make warning operations or directly block the call based on the call content. The intention recognition algorithm relies on training sets and fine-tuning of specific scenarios. In the context of the continuous updating of fraudulent rhetoric and the use of AI to commit fraud, there is a risk that the recognition rate and accuracy rate cannot adapt to the changes.

[0005] Based on the integrated modeling and analysis of the anti-fraud system and operator data, big data model training is carried out, and fraud risk prediction and measures are taken through situational awareness and real-time detection. This method relies on the sensitivity of algorithms and predictions to a certain extent, and there is a risk of being unable to identify and predict due to the accuracy of technical indicators and the variability of fraudulent behavior, resulting in low risk prediction recognition rate and accuracy. Summary of the invention

[0006] The purpose of this application is to provide an anti-fraud call method, device and equipment to solve the problem of low accuracy of current anti-fraud methods.

[0007] To achieve the above-mentioned purpose, an embodiment of the present application provides an anti-fraud call method, which is applied to a core service module, and the method includes:

[0008] The core service module receives a first request message sent by an application service (AS) module, wherein the first request message is used to request: response information of the AS module when simulating a virtual holder of a first number to talk with a first terminal;

[0009] Perform a first operation according to the call context information and the configuration information of the virtual holder;

[0010] The first operation includes at least one of the following:

[0011] Generate response information;

[0012] Fraud detection through semantic analysis;

[0013] Conduct data collection;

[0014] The first terminal is numbered.

[0015] Optionally, in a case where the first operation includes generating response information, performing the first operation according to the call context information and the configuration information of the virtual holder includes:

[0016] generating a first prompt word according to the call context information and the configuration information of the virtual holder, wherein the first prompt word is a keyword that matches the virtual holder and is related to the response information;

[0017] According to the call context information, a search is performed in a database using a Retrieval Augmented Generation (RAG) module to obtain search information, where the search information is a response content related to the call context information in the database;

[0018] generating a second prompt word according to the call context information, the first prompt word and the search information;

[0019] The second prompt word is input into a large language model (LLM) to obtain response information.

[0020] Optionally, the method further includes:

[0021] The operation data of the first operation is sent to the operation module, and the operation module is used to synchronize the operation data to the anti-fraud platform.

[0022] Optionally, the method further includes:

[0023] Obtaining the anti-fraud operation data of the anti-fraud platform for the call;

[0024] The LLM model is iteratively trained according to the operation data of the first operation and the anti-fraud operation data.

[0025] Optionally, the method further includes:

[0026] Receive the fourth request message sent by the AS module, where the fourth request message is used to request the agent session processing service.

[0027] To achieve the above object, an embodiment of the present application provides an anti-fraud call method, which is applied to an application service AS module, and the method includes:

[0028] When simulating a virtual holder of the first number to talk to the first terminal, the AS module sends a first request message to the core service module according to the configuration information of the virtual holder, where the first request message is used to request response information from the core service module;

[0029] receiving a response message sent by the core service module, wherein the response message is related to the configuration information of the virtual holder and the call context information;

[0030] Sending the response information to the first terminal.

[0031] Optionally, before sending the first request message to the core service module, the method further includes:

[0032] Sending a second request message to the media capability module, where the second request message carries voice call context information of a call between the AS module and the first terminal, and the second request message is used to request the media capability module to convert the voice call context information into text call context information;

[0033] Receive text call context information sent by the media capability module.

[0034] Optionally, before sending the response information to the first terminal, the method further includes:

[0035] Sending a third request message to the media capability module, the third request message carrying text response information, the third request message being used to request the media capability module to convert the text response information into voice response information;

[0036] The sending the response information to the first terminal includes:

[0037] The voice response information is sent to the first terminal.

[0038] Optionally, the first request message is further used to request the core service module to perform an anti-fraud related operation, where the anti-fraud related operation includes at least one of the following:

[0039] Fraud detection through semantic analysis;

[0040] Conduct data collection;

[0041] The first terminal is numbered.

[0042] Optionally, the method further includes:

[0043] When the call with the first terminal is connected, if the call meets the first condition, first indication information is sent to the core processing module, where the first indication information is used to instruct the core processing module to perform audio and video anchoring and media negotiation with the first terminal and the AS module;

[0044] The first condition includes at least one of the following:

[0045] The call is an audio or video call;

[0046] The terminal where the AS module is located supports audio and video calls;

[0047] The terminal where the AS module is located supports call service preprocessing.

[0048] Optionally, before sending the first request message to the core service module, the method further includes:

[0049] A fourth request message is sent to the core service module, where the fourth request message is used to request the agent session processing service.

[0050] Optionally, before sending the first request message to the core service module, the method further includes:

[0051] The configuration information of the virtual holder of the first number is obtained from the operation module, wherein the configuration information of the virtual holder of each number in the number set is stored in the operation module.

[0052] To achieve the above objectives, an embodiment of the present application provides an anti-fraud system, including:

[0053] The AS module is used to send a first request message to the core service module according to the configuration information of the virtual holder when simulating a virtual holder of the first number to talk with the first terminal, wherein the first request message is used to request response information from the core service module;

[0054] The core service module is used to receive the first request message sent by the AS module; and perform a first operation according to the call context information and the configuration information of the virtual holder;

[0055] The first operation includes at least one of the following:

[0056] Generate response information;

[0057] Fraud detection through semantic analysis;

[0058] Conduct data collection;

[0059] marking the first terminal with a number;

[0060] The AS module is also used for: receiving response information sent by the core service module; and sending the response information to the first terminal.

[0061] Optionally, the system further includes:

[0062] The media capability module is used to convert the voice call context information sent by the AS module into text call context information; and / or to convert the text response information sent by the AS module into voice response information.

[0063] Optionally, the system further includes:

[0064] The operation module is used to synchronize the operation data of the first operation to the anti-fraud platform.

[0065] To achieve the above-mentioned purpose, an embodiment of the present application provides an anti-fraud call device, comprising:

[0066] The core service module is used to receive a first request message sent by the application service AS module, wherein the first request message is used to request: response information of the AS module when simulating a virtual holder of the first number to talk with the first terminal;

[0067] Perform a first operation according to the call context information and the configuration information of the virtual holder;

[0068] The first operation includes at least one of the following:

[0069] Generate response information;

[0070] Fraud detection through semantic analysis;

[0071] Conduct data collection;

[0072] The first terminal is numbered.

[0073] To achieve the above-mentioned purpose, an embodiment of the present application provides an anti-fraud call device, comprising:

[0074] The AS module is used to send a first request message to the core service module according to the configuration information of the virtual holder when simulating a virtual holder of the first number to talk with the first terminal, wherein the first request message is used to request response information from the core service module;

[0075] receiving a response message sent by the core service module, wherein the response message is related to the configuration information of the virtual holder and the call context information;

[0076] Sending the response information to the first terminal.

[0077] To achieve the above-mentioned purpose, an embodiment of the present application provides a communication device, including a transceiver, a processor, a memory, and a program or instruction stored in the memory and executable on the processor; when the processor executes the program or instruction, the steps of the above-mentioned anti-fraud call method are implemented.

[0078] To achieve the above objectives, an embodiment of the present application provides a readable storage medium on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the anti-fraud call method as described above are implemented.

[0079] To achieve the above objectives, an embodiment of the present application provides a computer program product on which a program or instruction is stored, including computer instructions, which implement the steps of the above anti-fraud call method when executed by a processor.

[0080] The beneficial effects of the above technical solution of the present application are as follows:

[0081] In an embodiment of the present application, the AS module sends a first request message to the core service module to request response information when simulating a virtual holder of the first number to have a call with the first terminal. After receiving the first request message, the core service module generates response information based on the call context information and the configuration information of the virtual holder of the first number, and can also perform fraud detection, data collection, number labeling, etc. based on the call context information. By setting some numbers as honeypot numbers and configuring their corresponding virtual holders to simulate real people to communicate with other terminals, the latest suspected fraud calls can be obtained through active induction, and call content generation, fraud detection, data collection, number labeling, etc. can be performed in the process. Compared with the tracing process after the event occurs, this method can proactively identify and discover suspected fraud numbers in advance, and predict fraud risks and take measures at a lower cost and without causing actual losses, thereby improving the accuracy of fraud identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 This is one of the flowcharts of the anti-fraud call method according to an embodiment of the present application;

[0083] Figure 2 A schematic diagram of the structure of the anti-fraud system of an embodiment of the present application;

[0084] Figure 3 This is a second flow chart of the anti-fraud call method according to an embodiment of the present application;

[0085] Figure 4 This is a flowchart of the anti-fraud call method according to an embodiment of the present application;

[0086] Figure 5This is a fourth flowchart of the anti-fraud call method according to an embodiment of the present application;

[0087] Figure 6 This is a functional diagram of each module in the anti-fraud system of an embodiment of the present application;

[0088] Figure 7 This is one of the structural schematic diagrams of the communication device according to the embodiment of the present application;

[0089] Figure 8 This is a second structural diagram of a communication device according to an embodiment of the present application;

[0090] Fig. 9 This is the third structural diagram of the communication device according to the embodiment of the present application. DETAILED DESCRIPTION

[0091] In order to make the technical problems, technical solutions and advantages to be solved by the present application clearer, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0092] It should be understood that the references to "one embodiment" or "an embodiment" throughout the specification mean that the specific features, structures, or characteristics associated with the embodiment are included in at least one embodiment of the present application. Therefore, the references to "in one embodiment" or "in an embodiment" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0093] In the various embodiments of the present application, it should be understood that the size of the serial numbers of the following processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0094] Additionally, the terms "system" and "network" are often used interchangeably herein.

[0095] In the embodiments provided in the present application, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined according to A. However, it should also be understood that determining B according to A does not mean determining B only according to A, and B can also be determined according to A and / or other information.

[0096] like Figure 1 As shown, the embodiment of the present application provides an anti-fraud call method, which is applied to the core service module, including:

[0097] Step 101: The core service module receives a first request message sent by the application service AS module, where the first request message is used to request: response information of the AS module when simulating a virtual holder of a first number to talk with a first terminal;

[0098] Perform a first operation according to the call context information and the configuration information of the virtual holder;

[0099] The first operation includes at least one of the following:

[0100] Generate response information;

[0101] Fraud detection through semantic analysis;

[0102] Conduct data collection;

[0103] The first terminal is numbered.

[0104] In this embodiment, the core service module is the core service part of the anti-fraud system, and the core service module can provide artificial intelligence (AI) agent service, RAG service and anti-fraud LLM service. Figure 2 As shown, the anti-fraud system of the embodiment of the present application also includes: an AS module, which is a control plane network element in the anti-fraud system; a core processing module, which is a core network element / platform in the system, including but not limited to Voice over Long-Term Evolution (VOLTE) AS, Voice Over Network Routing (VoNR) + capability network element, VoNR + media plane; a media capability module, which is a media processing core platform in the anti-fraud system; an operation module, which is used to provide a management configuration component for the service. The anti-fraud system also includes: a number set, which is stored in a number pool, and the number set includes one or more numbers with corresponding virtual holders. The operation module performs virtual settings for different numbers and stores the configuration information of each virtual holder, and the configuration information includes but is not limited to at least one of the following: the holder's name, age, gender, place of origin, occupation, address, accent, ID number, personality traits, consumption tendency, financial management tendency, risk awareness, past experience, etc.

[0105] like Figure 2 As shown, each module in the anti-fraud system interacts with the external terminal through the core processing module to realize the call between the virtual holder and the external terminal. Each module in the anti-fraud system interacts with the anti-fraud platform through the operation module, for example, the operation module sends the data of the call process to the anti-fraud platform, and the anti-fraud platform sends the processed data of the call to the operation module.

[0106] The AS module can simulate the virtual holder of the first number to conduct a conversation with the first terminal, and the call can be a calling call or a called call. For example, the AS module simulates virtual holders of different numbers to make active outbound calls to suspected fraudulent numbers posted on forums and websites according to a plan; or, the AS module intelligently answers the called party's call, for example, for various scenarios such as impersonating an acquaintance, impersonating a public security, procuratorial or judicial organ, impersonating a bank employee, impersonating an online customer service, etc., and simulates a real person interacting with the caller in accordance with the settings of the virtual holder of the number.

[0107] When the AS module is in a call with the first terminal (for example, when the call is picked up on behalf of the first terminal), the AS module sends a first request message to the core service module to request response information of the call with the first terminal. The first request message carries the call context information between the AS module and the first terminal, and the call context information is, for example, interactive information such as the call information sent by the first terminal and / or the response information replied by the AS module.

[0108] After receiving the first request message, the core service module calls the LLM model, RAG, agent service and other capabilities during the processing to generate response information, perform fraud detection, data collection, number labeling, etc. The core service module sends the generated response information to the AS module, and the AS module sends the response information to the first terminal to communicate with the first terminal.

[0109] In an embodiment of the present application, the AS module sends a first request message to the core service module to request a response message when simulating a virtual holder of the first number to talk to the first terminal. After receiving the first request message, the core service module generates a response message based on the call context information and the configuration information of the virtual holder of the first number, and can also perform fraud detection, data collection, number labeling, etc. based on the call context information. By setting some numbers as honeypot numbers and configuring their corresponding virtual holders to simulate real people to communicate with other terminals, the latest suspected fraud calls can be obtained through active induction, and call content generation, fraud detection, data collection, number labeling, etc. can be performed in the process. Compared with the tracing process after the incident occurs, this method can proactively identify and discover suspected fraud numbers in advance, and predict fraud risks and take measures at a lower cost and without causing actual losses, thereby improving the accuracy of fraud identification. Improve the user experience of online users, improve user satisfaction, and reduce complaints caused by fraud, harassment, etc.

[0110] As an optional embodiment, the method further includes:

[0111] Receive the fourth request message sent by the AS module, where the fourth request message is used to request the agent session processing service.

[0112] In this embodiment, before the AS module simulates a real person talking to the first terminal and after the AS module picks up the phone, it first needs to request the agent session processing service from the operation module, that is, request each agent in the operation module to process the current session.

[0113] As an optional embodiment, in the case where the first operation includes generating response information, performing the first operation according to the call context information and the configuration information of the virtual holder includes:

[0114] generating a first prompt word according to the call context information and the configuration information of the virtual holder, wherein the first prompt word is a keyword that matches the virtual holder and is related to the response information;

[0115] According to the call context information, a RAG module is used to search in a database to obtain search information, where the search information is a response content related to the call context information in the database;

[0116] generating a second prompt word according to the call context information, the first prompt word and the search information;

[0117] The second prompt word is input into the large language model LLM to obtain response information.

[0118] In this embodiment, the core service module may include one or more service units such as an agent service unit, a RAG module, an LLM model, a fraud detection agent, a data collection agent, and other agents. Different units have different processing functions. For example, the RAG module can retrieve the response words and content in the database, the LLM model can generate the response words based on the input data information, the fraud detection agent can implement fraud detection, and the data collection agent can be used to collect data. Taking the call received from the first terminal as an example, the steps for the core service module to generate the response information are as follows: Figure 3 As shown, including:

[0119] Step 31, the agent service of the core service module performs semantic segmentation on the received call content (call context information), and generates a prompt word (Prompt) in combination with the current virtual holder setting (determined according to the configuration information) and the call context content, which may also be called the first prompt word;

[0120] Step 32: According to the answering strategy and the collaborative arrangement of other agents, the call context information is synchronized to each agent. Each agent performs relevant processing according to its own functional settings, such as fraud detection, data collection, number labeling and other tasks. During the execution of tasks, each agent may also involve interaction with the LLM model and RAG module. The interaction process is implemented according to the functions of each agent.

[0121] The answering strategy is pre-set and is related to the configuration information of the virtual holder and the scenario of the call. The call scenarios may include product marketing, imitation of relatives and friends, fake order rebate, false credit investigation, etc. For different call scenarios, the response information of the virtual holder may be different; for virtual holders with different configurations (such as different personalities, different work experiences, and different ages), the response information for this call may also be different. Therefore, different answering strategies are set in advance for various types of call scenarios and virtual holders.

[0122] Step 33: The pick-up Agent service sends the call context content to the RAG module for content enhancement retrieval.

[0123] Step 34 to Step 35: After the RAG module vectorizes the call context, it searches the internal vector database and obtains the search information of the response words and the response content, and returns the search information to the pick-up Agent service.

[0124] Step 36: The pick-up Agent service combines the call context, the search information and the original prompt word Prompt (the first prompt word) to generate an enhanced prompt word Prompt, which may also be called a second prompt word.

[0125] Step 37 to Step 38: By using the enhanced prompt word Prompt, the LLM model is requested to generate a response script, and the LLM model returns the response information.

[0126] As an optional embodiment, the method further includes:

[0127] The operation data of the first operation is sent to the operation module, and the operation module is used to synchronize the operation data to the anti-fraud platform.

[0128] In this embodiment, after the core service module performs the first operation, the relevant operation data can be sent to the operation module, and the operation module can send the operation data to the anti-fraud platform. For example, the operation module coordinates the operation data and results of the called party access or outbound call task with the anti-fraud platform according to the strategy, and the anti-fraud platform performs the next anti-fraud operation in a targeted manner. This part of the work consists of two parts:

[0129] Data collection, number labeling, fraud detection, and response information generated in the core service module are synchronized to the operation module;

[0130] The operation module synchronizes data and triggers functions with the anti-fraud platform according to pre-made strategies, and the anti-fraud platform performs specific anti-fraud operations on the data. The anti-fraud operations include recording and tracking the number of the first terminal, intercepting numbers related to the first number, generating warnings, etc.

[0131] As an optional embodiment, the method further includes:

[0132] Obtain anti-fraud operation data of the anti-fraud platform for the call; and iteratively train the LLM model according to the operation data of the first operation and the anti-fraud operation data.

[0133] In this embodiment, the anti-fraud platform can synchronize relevant operation data to the operation module, and the operation module sends the anti-fraud operation data to the core service module, or the core service module obtains the anti-fraud operation data from the operation module, and iteratively trains and optimizes the LLM model based on the anti-fraud operation data, call data, etc. For example: the core service module integrates the called access or outbound call task process data and result feedback, as well as the feedback data of the anti-fraud platform, and iteratively trains and optimizes the anti-fraud model (LLM model). This part of the work consists of two parts:

[0134] The anti-fraud platform synchronizes the processing results of corresponding events to the operation module, forming an end-to-end data closed loop.

[0135] The anti-fraud model in the core service module is iteratively trained and optimized based on the latest data.

[0136] In this embodiment, AI technologies such as anti-fraud models, RAG, and AI Agent are important applications of AI+ security empowerment and have great versatility. Through continuous iteration and optimization of the model and continuous improvement of information in the RAG plug-in database, the anti-fraud efficiency can be greatly improved and labor costs can be reduced.

[0137] Based on this method, outbound calls can be made automatically and intelligently, and a real person can be simulated to interact with the called party. For example, different numbers can be scheduled to make active outbound calls to suspected fraud numbers posted on forums and websites, and semantic and intent recognition and number marking can be performed through the content in the communication process; intelligent answering can be performed, and in various scenarios such as impersonating acquaintances, impersonating public security, procuratorial and judicial organs, impersonating bank staff, and impersonating online customer service, according to the settings of the virtual holder of the number, a real person can be simulated to interact with the caller, and semantic and intent recognition and number marking can be performed through the content in the communication process, and further anti-fraud follow-up operations such as code number marking can be performed.

[0138] The following examples illustrate the implementation steps of the anti-fraud call method of this application:

[0139] Preliminary preparation: including the management of numbers in the number pool and the setting of virtual holders for different numbers through the operation module, including but not limited to the holder's name, age, gender, place of origin, occupation, address, accent, ID number, personality traits, consumption tendencies, financial management tendencies, risk awareness, past experience, etc.

[0140] Step 1: The AS module takes over the number's called access according to the number's virtual holder settings. The AS module calls the core service module to simulate a real person's call communication, and performs call content generation, fraud detection, data collection, number labeling, etc. in the process.

[0141] Specifically, the AS module calls the core service module to simulate the steps of real-person communication. Figure 4 As shown, including:

[0142] Step 41: Terminal A initiates a call request to call a number in a preset number set.

[0143] Step 42: The core processing module sends a call event notification; the AS module returns a confirmation message.

[0144] Step 43: The AS module goes off-hook and the session is connected.

[0145] Step 44: The AS module identifies whether the service triggering conditions are currently met (such as audio and video calls, whether the terminal supports audio and video calls, and call service preprocessing), and issues a call event control to instruct two-way audio and video anchoring;

[0146] Step 45: Terminal A, the core processing module, and the AS module complete the audio and video anchoring according to the service triggering process.

[0147] Step 46: Complete the audio and video media negotiation process.

[0148] Step 47: The AS module obtains the configuration information of the virtual holder corresponding to the current number from the operation module.

[0149] Step 48: The AS module requests the core service module to connect to the Agent service to process the current session.

[0150] Step 49: The AS module requests the media capability module for a speech to text (STT) service (converting the speech during the call into text).

[0151] Step 410: The media capability module performs scheduling negotiation internally, executes the speech-to-text service, and then returns the processing result address to the AS module.

[0152] Step 411: The AS module requests the core service module to receive the response information of the Agent service, and the request message carries the speech-to-text service address.

[0153] Step 412: The core service module calls the LLM model, RAG and other capabilities during the processing, and can asynchronously perform tasks such as AIGC response, fraud detection, data collection, number detection, etc. The execution process of the core service module generating a response message, performing fraud detection, data collection, number detection and other tasks will not be described in detail here.

[0154] Step 413: The AS module requests a text to speech (TTS) service from the media capability module.

[0155] Step 414: The media capability module performs scheduling negotiation internally, executes the text-to-speech service, and returns the processing result address to the AS module.

[0156] Step 415: the AS module negotiates with the core processing module network on "audio and video media stream update", and the media stream after text-to-speech processing is sent to terminal A through the core processing module.

[0157] Step 2: The AS module calls the core service module to simulate a real person's call communication, and continuously generates call content, detects fraud, collects data, and labels numbers during the call. This part of the work is implemented by the core service module through the arrangement and calling of different functional modules.

[0158] Step 3: The operation module coordinates the data and results of the caller access or outbound call task with the anti-fraud platform according to the strategy, and the anti-fraud platform performs the next anti-fraud operation in a targeted manner. This part of the work consists of two parts:

[0159] The data collection, number labeling and other units in the core service module synchronize data to the operation module;

[0160] The operation module synchronizes data and triggers functions with the anti-fraud platform based on pre-made strategies, and the anti-fraud platform performs specific anti-fraud operations based on the data.

[0161] Step 4: The core service module integrates the process data and result feedback of the called party access or outbound call task, as well as the feedback data of the anti-fraud platform, and iteratively trains and optimizes the anti-fraud model (LLM model). This part of the work consists of two parts:

[0162] The anti-fraud platform synchronizes the processing results of corresponding events to the operation module, forming an end-to-end data closed loop.

[0163] The anti-fraud model in the core service module is iteratively trained and optimized based on the latest data.

[0164] In this embodiment, based on proactive services in advance, the source of fraud is blocked in advance, and the effect is significantly better than post-tracing processing; based on AI technologies such as anti-fraud models, RAG, and AI Agent (intelligent body), it is an important application of AI+ security empowerment, with great versatility. Through the continuous iteration and optimization of large models and the continuous improvement of information in the RAG plug-in database, it will greatly improve the anti-fraud efficiency and reduce labor costs. This solution can be integrated and collaborated with existing anti-fraud platforms, providing the anti-fraud platform with a large amount of suspected fraud data and clues, and organically combined with the existing big data trend analysis and anti-fraud model strategy analysis of the anti-fraud platform, to improve the detection accuracy and conduct intelligent perception and discovery of gang-based fraud; at the same time, the final case processing data of the anti-fraud platform is used for model training, which can effectively promote the iterative training and optimization of large models and form an anti-fraud business closed loop.

[0165] This solution has a wide range of application scenarios in the protection, prevention and detection of telecommunications network fraud. Based on this solution, suspected fraud codes can be identified in advance, the user experience of online users can be improved, user satisfaction can be improved, and complaints caused by fraud, harassment, etc. can be reduced. It has a great impact on improving the corporate social responsibility image and social influence, and indirectly brings business improvement and user conversion. The embodiments of the present application can also be applied to a wider range of telecommunications fraud scenarios other than calls, and can be integrated in more scenarios such as websites, applets, and APP applications to play a greater value.

[0166] In this application, the numbers in the number set can be the "secondary numbers" of the operator, such as numbers that have been cancelled or idle. Combined with the artificial intelligence generated content (AIGC) technology of the big model, a "number honeypot" for people at high risk of fraud is constructed. The latest suspected fraud calls are obtained through active inducement, and the words and behavioral data of the suspected fraudsters in the "honeypot" are used to feed back to the big model and algorithm, and the relevant information is reported to the anti-fraud platform and public security systems. At a very low cost and without causing actual losses, the fraud risk can be predictively handled and measures can be taken.

[0167] like Figure 5 As shown, the embodiment of the present application also provides an anti-fraud call method, which is applied to the application service AS module, including:

[0168] Step 501: When simulating a virtual holder of a first number to talk to a first terminal, the AS module sends a first request message to a core service module according to configuration information of the virtual holder, where the first request message is used to request response information from the core service module.

[0169] Step 502: receiving a response message sent by the core service module, wherein the response message is related to the configuration information of the virtual holder and the call context information;

[0170] Step 503: Send the response information to the first terminal.

[0171] In this embodiment, the core service module is the core service part of the anti-fraud system, and the core service module can provide AI Agent service, RAG service and anti-fraud LLM and service. The AS module is the control plane network element in the anti-fraud system. The anti-fraud system of the embodiment of the present application also includes: a core processing module, which is the core network element / platform in the system, including but not limited to VOLTE AS, VoNR+ capability network element, and VoNR+ media plane; a media capability module, which is the media processing core platform in the anti-fraud system; an operation module, which is used to provide a management configuration component for the service. The anti-fraud system also includes: a number set, which is stored in a number pool, and the number set includes one or more numbers with corresponding virtual holders. The operation module performs virtual settings for different numbers and stores the configuration information of each virtual holder, and the configuration information includes but is not limited to at least one of the following: holder's name, age, gender, native place, occupation, address, accent, ID number, personality traits, consumption tendency, financial management tendency, risk awareness, past experience, etc.

[0172] The AS module can simulate the virtual holder of the first number to conduct a conversation with the first terminal, and the call can be a calling call or a called call. For example, the AS module simulates virtual holders of different numbers to make active outbound calls to suspected fraudulent numbers posted on forums and websites according to a plan; or, the AS module intelligently answers the called party's call, for example, for various scenarios such as impersonating an acquaintance, impersonating a public security, procuratorial or judicial organ, impersonating a bank employee, impersonating an online customer service, etc., and simulates a real person interacting with the caller in accordance with the settings of the virtual holder of the number.

[0173] When the AS module is in a call with the first terminal (for example, when the call is picked up on behalf of the first terminal), the AS module sends a first request message to the core service module to request response information of the call with the first terminal. The first request message carries the call context information between the AS module and the first terminal, and the call context information is, for example, interactive information such as the call information sent by the first terminal and / or the response information replied by the AS module.

[0174] After receiving the first request message, the core service module calls the LLM model, RAG, agent service and other capabilities during the processing to generate response information, perform fraud detection, data collection, number detection, etc. The core service module sends the generated response information to the AS module, and the AS module sends the response information to the first terminal to communicate with the first terminal.

[0175] In an embodiment of the present application, the AS module sends a first request message to the core service module to request response information when simulating a call between a virtual holder of a first number and a first terminal. After receiving the first request message, the core service module generates response information based on the call context information and the configuration information of the virtual holder of the first number. It can also perform fraud detection, data collection, number labeling, etc. based on the call context information. This method can proactively identify and discover suspected fraudulent numbers in advance, and predictively handle fraud risks and take measures at a lower cost and without causing actual losses, thereby improving the accuracy of fraud identification. Improve the user experience of online users, improve user satisfaction, and reduce complaints caused by fraud, harassment, etc.

[0176] As an optional embodiment, before sending the first request message to the core service module, the method further includes:

[0177] A second request message is sent to the media capability module, where the second request message carries the voice call context information of the call between the AS module and the first terminal, and the second request message is used to request the media capability module to convert the voice call context information into text call context information; and the text call context information sent by the media capability module is received.

[0178] In this embodiment, after the AS module picks up the phone, it can request the media capability module to perform a voice-to-text service, such as converting the voice message sent by the first terminal into a text message. The text message is used to execute a subsequent response request, such as sending a first request message to the core service module, and the first request message carries the call context information in text form. The response information returned by the core service module is also a text-based response information.

[0179] Optionally, before sending the response information to the first terminal, the method further includes:

[0180] Sending a third request message to the media capability module, the third request message carrying text response information, the third request message being used to request the media capability module to convert the text response information into voice response information;

[0181] The sending the response information to the first terminal includes: sending the voice response information to the first terminal.

[0182] In this embodiment, since the response information returned by the core service module and received by the AS module is in text form, the AS module sends the text-form response information to the media capability module, so that the media capability module performs a text-to-speech service, thereby obtaining the voice-form response information, and sends the voice-form response information to the first terminal through the core processing module.

[0183] As an optional embodiment, the first request message is further used to request the core service module to perform an anti-fraud related operation, and the anti-fraud related operation includes at least one of the following:

[0184] Fraud detection through semantic analysis;

[0185] Conduct data collection;

[0186] The first terminal is numbered.

[0187] In this embodiment, the AS module sends a first request message to the core service module while simulating the virtual holder of the first number to have a call with the first terminal. After receiving the first request message, the core service module generates a response message and can also perform fraud detection, data collection, number labeling, etc. based on internal LLM models, RAGs, intelligent agent services and other capabilities.

[0188] As an optional embodiment, the method further includes:

[0189] When the call with the first terminal is connected, if the call meets the first condition, first indication information is sent to the core processing module, where the first indication information is used to instruct the core processing module to perform audio and video anchoring and media negotiation with the first terminal and the AS module;

[0190] The first condition includes at least one of the following:

[0191] The call is an audio or video call;

[0192] The terminal where the AS module is located supports audio and video calls;

[0193] The terminal where the AS module is located supports call service preprocessing.

[0194] In this embodiment, when the AS module receives a call request from the first terminal and picks up the phone, it identifies whether the service triggering conditions (such as audio and video calls, whether the terminal supports audio and video calls, call service preprocessing, etc.) are currently met. If the triggering conditions are met, a call event control is issued to instruct to perform two-way audio and video anchoring and media negotiation. The first terminal, the core processing module, and the AS module complete the audio and video anchoring and media negotiation process according to the service triggering process.

[0195] Optionally, before sending the first request message to the core service module, the method further includes:

[0196] A fourth request message is sent to the core service module, where the fourth request message is used to request the agent session processing service.

[0197] In this embodiment, before the AS module simulates a real person talking to the first terminal and after the AS module picks up the phone, it first needs to request the agent session processing service from the operation module, that is, request each agent in the operation module to process the current session.

[0198] Optionally, before sending the first request message to the core service module, the method further includes:

[0199] The configuration information of the virtual holder of the first number is obtained from the operation module, wherein the configuration information of the virtual holder of each number in the number set is stored in the operation module.

[0200] In this embodiment, the operation module stores configuration information of virtual holders corresponding to multiple numbers. The first number is a number for communicating with the first terminal. When the AS module simulates a real person talking to the first terminal, it needs to first obtain the configuration information of the virtual holder of the first number, and obtain response information matching the virtual holder based on the configuration information.

[0201] The following examples illustrate the implementation steps of the anti-fraud call method of this application:

[0202] Preliminary preparation: including the management of numbers in the number pool and the setting of virtual holders for different numbers through the operation module, including but not limited to the holder's name, age, gender, place of origin, occupation, address, accent, ID number, personality traits, consumption tendencies, financial management tendencies, risk awareness, past experience, etc.

[0203] Step 1: The AS module takes over the number's called access according to the number's virtual holder settings. The AS module calls the core service module to simulate a real person's call communication, and performs call content generation, fraud detection, data collection, number labeling, etc. in the process.

[0204] Specifically, the AS module calls the core service module to simulate the steps of real-person communication. Figure 4 As shown, including:

[0205] Step 41: Terminal A initiates a call request to call a number in a preset number set.

[0206] Step 42: The core processing module sends a call event notification; the AS module returns a confirmation message.

[0207] Step 43: The AS module goes off-hook and the session is connected.

[0208] Step 44: The AS module identifies whether the service triggering conditions are currently met (such as audio and video calls, whether the terminal supports audio and video calls, and call service preprocessing), and issues a call event control to instruct two-way audio and video anchoring;

[0209] Step 45: Terminal A, the core processing module, and the AS module complete the audio and video anchoring according to the service triggering process.

[0210] Step 46: Complete the audio and video media negotiation process.

[0211] Step 47: The AS module obtains the configuration information of the virtual holder corresponding to the current number from the operation module.

[0212] Step 48: The AS module requests the core service module to connect to the Agent service to process the current session.

[0213] Step 49: The AS module requests the media capability module to provide a speech-to-text service (converting the speech during the call into text).

[0214] Step 410: The media capability module performs scheduling negotiation internally, executes the speech-to-text service, and then returns the processing result address to the AS module.

[0215] Step 411: The AS module requests the core service module to receive the response information of the Agent service, and the request message carries the speech-to-text service address.

[0216] Step 412: The core service module calls the LLM model, RAG and other capabilities during the processing, and can asynchronously perform tasks such as AIGC response, fraud detection, data collection, number detection, etc. The execution process of the core service module generating a response message, performing fraud detection, data collection, number detection and other tasks will not be described in detail here.

[0217] Step 413: The AS module requests the media capability module for a text-to-speech service.

[0218] Step 414: The media capability module performs scheduling negotiation internally, executes the text-to-speech service, and returns the processing result address to the AS module.

[0219] Step 415: the AS module negotiates with the core processing module network on "audio and video media stream update", and the media stream after text-to-speech processing is sent to terminal A through the core processing module.

[0220] Step 2: The AS module calls the core service module to simulate a real person's call communication, and continuously generates call content, detects fraud, collects data, and labels numbers during the call. This part of the work is implemented by the core service module through the arrangement and calling of different functional modules.

[0221] Step 3: The operation module coordinates the data and results of the caller access or outbound call task with the anti-fraud platform according to the strategy, and the anti-fraud platform performs the next anti-fraud operation in a targeted manner. This part of the work consists of two parts:

[0222] The data collection, number labeling and other units in the core service module synchronize data to the operation module;

[0223] The operation module synchronizes data and triggers functions with the anti-fraud platform based on pre-made strategies, and the anti-fraud platform performs specific anti-fraud operations based on the data.

[0224] Step 4: The core service module integrates the process data and result feedback of the called party access or outbound call task, as well as the feedback data of the anti-fraud platform, and iteratively trains and optimizes the anti-fraud model (LLM model). This part of the work consists of two parts:

[0225] The anti-fraud platform synchronizes the processing results of corresponding events to the operation module, forming an end-to-end data closed loop.

[0226] The anti-fraud model in the core service module is iteratively trained and optimized based on the latest data.

[0227] In an embodiment of the present application, the AS module sends a first request message to the core service module to request response information when simulating a call between a virtual holder of a first number and a first terminal. After receiving the first request message, the core service module generates response information based on the call context information and the configuration information of the virtual holder of the first number. It can also perform fraud detection, data collection, number labeling, etc. based on the call context information. This method can proactively identify and discover suspected fraudulent numbers in advance, and predictively handle fraud risks and take measures at a lower cost and without causing actual losses, thereby improving the accuracy of fraud identification. Improve the user experience of online users, improve user satisfaction, and reduce complaints caused by fraud, harassment, etc.

[0228] The embodiment of the present application also provides an anti-fraud system, such as Figure 2 As shown, including:

[0229] The AS module is used to send a first request message to the core service module according to the configuration information of the virtual holder when simulating a virtual holder of the first number to talk with the first terminal, wherein the first request message is used to request response information from the core service module;

[0230] The core service module is used to receive the first request message sent by the AS module; and perform a first operation according to the call context information and the configuration information of the virtual holder;

[0231] The first operation includes at least one of the following:

[0232] Generate response information;

[0233] Fraud detection through semantic analysis;

[0234] Conduct data collection;

[0235] marking the first terminal with a number;

[0236] The AS module is also used for: receiving response information sent by the core service module; and sending the response information to the first terminal.

[0237] The core service module is the core service part of the anti-fraud system, and the core service module can provide AI Agent service, RAG service and anti-fraud LLM service. The AS module is the control plane network element in the anti-fraud system;

[0238] Optionally, the AS module is further used to: when the call with the first terminal is connected, if the call meets the first condition, send first indication information to the core processing module, wherein the first indication information is used to instruct the core processing module to perform audio and video anchoring and media negotiation with the first terminal and the AS module;

[0239] The first condition includes at least one of the following:

[0240] The call is an audio or video call;

[0241] The terminal where the AS module is located supports audio and video calls;

[0242] The terminal where the AS module is located supports call service preprocessing.

[0243] Optionally, the AS module is further used to: send a fourth request message to the core service module, where the fourth request message is used to request an agent session processing service.

[0244] Optionally, the core service module is also used to: receive a fourth request message sent by the AS module, where the fourth request message is used to request an agent session processing service.

[0245] Optional, such as Figure 2 As shown, the system further includes: an operation module;

[0246] The AS module is further used to obtain configuration information of the virtual holder of the first number from the operation module, wherein the configuration information of the virtual holder of each number in the number set is stored in the operation module.

[0247] Optionally, the core service module is further used to: generate a first prompt word according to the call context information and the configuration information of the virtual holder, wherein the first prompt word is a keyword that matches the virtual holder and is related to the response information;

[0248] According to the call context information, a RAG module is used to search in a database to obtain search information, where the search information is a response content related to the call context information in the database;

[0249] generating a second prompt word according to the call context information, the first prompt word and the search information;

[0250] The second prompt word is input into the large language model LLM to obtain response information.

[0251] Optionally, the core service module is further used to: send the operation data of the first operation to the operation module, and the operation module is used to synchronize the operation data to the anti-fraud platform.

[0252] like Figure 2 As shown, the system further includes: a media capability module, which is used to convert the voice call context information sent by the AS module into text call context information; and / or, which is used to convert the text response information sent by the AS module into voice response information. The media capability module is the core media processing platform in the anti-fraud system.

[0253] Optionally, the AS module is further used to: send a second request message to the media capability module, the second request message carries voice call context information of the call between the AS module and the first terminal, and the second request message is used to request the media capability module to convert the voice call context information into text call context information. The media capability module is used to perform a voice-to-text service and send the converted text information to the AS module.

[0254] Optionally, the AS module is further used to: send a third request message to the media capability module, the third request message carries text response information, and the third request message is used to request the media capability module to convert the text response information into voice response information. The media capability module is also used to perform text-to-speech service and send the converted voice information to the AS module. The AS module sends the response information in voice form to the first terminal.

[0255] Optionally, the core service module is further used to: obtain anti-fraud operation data of the anti-fraud platform for the call; and iteratively train the LLM model according to the operation data of the first operation and the anti-fraud operation data.

[0256] like Figure 2 As shown, the anti-fraud system also includes:

[0257] The core processing module is the core network element / platform in the system, including but not limited to VOLTE AS, VoNR+ capability network element, and VoNR+ media plane; the media capability module is the core media processing platform in the anti-fraud system;

[0258] The number set is stored in the number pool. The number set includes one or more numbers with corresponding virtual holders. The operation module performs virtual settings on the virtual holders of different numbers and stores the configuration information of each virtual holder.

[0259] The functions supported by each module in the anti-fraud system are as follows: Figure 6 As shown, the AS module can realize control signaling docking and control information interaction among various modules according to the business execution logic. The AS module can communicate with other terminals through the core processing module, which can realize operations such as media stream playback, media stream copying and forwarding, and realize the transmission of call information between other terminals and the AS module.

[0260] The media capability module can perform text-to-speech and speech-to-text services at the request of the AS module, such as converting voice call information sent by other terminals into text call information, converting text response information generated by the core service module into voice response information, etc.

[0261] The core service module has functions such as intelligent agent management, anti-fraud model (LLM model) reasoning, RAG enhanced retrieval service, etc. It can generate response information, collect data, detect fraud, and label numbers.

[0262] The operation module can realize: management of number sets, such as number management in number sets, identity management of virtual holders, and management of number sets; management of data, such as management of annotation data, event data, virtual holder related data and other data; management of answering strategies, such as definition of each virtual holder role, different call scenarios and corresponding response strategies for different virtual holders.

[0263] The operation module can also exchange data with the anti-fraud platform, such as synchronizing relevant data during calls and synchronizing the anti-fraud operation data of the anti-fraud platform.

[0264] The anti-fraud system also includes other service modules, such as a service module for executing anti-fraud model training, a service module for generating virtual holder identities, an internal vector database, etc., which will not be elaborated here.

[0265] The anti-fraud system of the embodiment of the present application combines the AIGC technology of the big model to build a "number honeypot" for high-risk groups of fraud, obtains the latest suspected fraud calls through active inducement, feeds back the big model and algorithm based on the words and behavior data of the suspected fraudsters in the "honeypot", and reports relevant information to the anti-fraud platform and public security systems, so as to predict and deal with fraud risks and take measures at a very low cost and without causing actual losses. It can be integrated and collaborated with the existing anti-fraud platform to provide a large amount of suspected fraud data and clues to the anti-fraud platform, and organically combined with the existing big data trend analysis and anti-fraud model strategy analysis of the anti-fraud platform to improve the detection accuracy and conduct intelligent perception and discovery of gang-based fraud; at the same time, the final case processing data of the anti-fraud platform is used for model training, which can effectively promote the iterative training and optimization of the big model and form an anti-fraud business closed loop.

[0266] The embodiment of the present application also provides an anti-fraud call device, including:

[0267] The core service module is used to receive a first request message sent by the application service AS module, wherein the first request message is used to request: response information of the AS module when simulating a virtual holder of the first number to talk with the first terminal;

[0268] Perform a first operation according to the call context information and the configuration information of the virtual holder;

[0269] The first operation includes at least one of the following:

[0270] Generate response information;

[0271] Fraud detection through semantic analysis;

[0272] Conduct data collection;

[0273] The first terminal is numbered.

[0274] Optionally, the core service module is specifically used for:

[0275] generating a first prompt word according to the call context information and the configuration information of the virtual holder, wherein the first prompt word is a keyword that matches the virtual holder and is related to the response information;

[0276] According to the call context information, a RAG module is used to search in a database to obtain search information, where the search information is a response content related to the call context information in the database;

[0277] generating a second prompt word according to the call context information, the first prompt word and the search information;

[0278] The second prompt word is input into the large language model LLM to obtain response information.

[0279] Optionally, the core service module is also used to:

[0280] The operation data of the first operation is sent to the operation module, and the operation module is used to synchronize the operation data to the anti-fraud platform.

[0281] Optionally, the core service module is also used to:

[0282] Obtaining the anti-fraud operation data of the anti-fraud platform for the call;

[0283] The LLM model is iteratively trained according to the operation data of the first operation and the anti-fraud operation data.

[0284] Optionally, the core service module is also used to:

[0285] Receive the fourth request message sent by the AS module, where the fourth request message is used to request the agent session processing service.

[0286] It should be noted that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment applied to the core service module, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0287] The embodiment of the present application also provides an anti-fraud call device, including:

[0288] The AS module is used to send a first request message to the core service module according to the configuration information of the virtual holder when simulating a virtual holder of the first number to talk with the first terminal, wherein the first request message is used to request response information from the core service module;

[0289] receiving a response message sent by the core service module, wherein the response message is related to the configuration information of the virtual holder and the call context information;

[0290] Sending the response information to the first terminal.

[0291] Optionally, the AS module is further used for:

[0292] Sending a second request message to the media capability module, where the second request message carries voice call context information of a call between the AS module and the first terminal, and the second request message is used to request the media capability module to convert the voice call context information into text call context information;

[0293] Receive text call context information sent by the media capability module.

[0294] Optionally, the AS module is further used for:

[0295] Sending a third request message to the media capability module, the third request message carrying text response information, the third request message being used to request the media capability module to convert the text response information into voice response information;

[0296] The sending the response information to the first terminal includes:

[0297] The voice response information is sent to the first terminal.

[0298] Optionally, the first request message is further used to request the core service module to perform an anti-fraud related operation, where the anti-fraud related operation includes at least one of the following:

[0299] Fraud detection through semantic analysis;

[0300] Conduct data collection;

[0301] The first terminal is numbered.

[0302] Optionally, the AS module is further used for:

[0303] When the call with the first terminal is connected, if the call meets the first condition, first indication information is sent to the core processing module, where the first indication information is used to instruct the core processing module to perform audio and video anchoring and media negotiation with the first terminal and the AS module;

[0304] The first condition includes at least one of the following:

[0305] The call is an audio or video call;

[0306] The terminal where the AS module is located supports audio and video calls;

[0307] The terminal where the AS module is located supports call service preprocessing.

[0308] Optionally, the AS module is further used for:

[0309] A fourth request message is sent to the core service module, where the fourth request message is used to request the agent session processing service.

[0310] Optionally, the AS module is further used for:

[0311] The configuration information of the virtual holder of the first number is obtained from the operation module, wherein the configuration information of the virtual holder of each number in the number set is stored in the operation module.

[0312] It should be noted that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment applied to the AS module, and can achieve the same technical effect. The parts and beneficial effects that are the same as the method embodiment in this embodiment will not be described in detail here.

[0313] like Figure 7 As shown, a communication device in an embodiment of the present application, the communication device is a first communication device 700, the first communication device may be a core service module, including a processor 710 and a transceiver 720;

[0314] The transceiver 720 is used to: receive a first request message sent by the application service AS module, wherein the first request message is used to request: response information of the AS module when simulating a virtual holder of the first number to talk with the first terminal;

[0315] The processor 710 is used to: perform a first operation according to the call context information and the configuration information of the virtual holder;

[0316] The first operation includes at least one of the following:

[0317] Generate response information;

[0318] Fraud detection through semantic analysis;

[0319] Conduct data collection;

[0320] The first terminal is numbered.

[0321] Optionally, the processor 710 is specifically configured to:

[0322] generating a first prompt word according to the call context information and the configuration information of the virtual holder, wherein the first prompt word is a keyword that matches the virtual holder and is related to the response information;

[0323] According to the call context information, a search is performed in a database using a RAG enhanced search module to obtain search information, where the search information is a response content related to the call context information in the database;

[0324] generating a second prompt word according to the call context information, the first prompt word and the search information;

[0325] The second prompt word is input into the large language model LLM to obtain response information.

[0326] Optionally, the transceiver 720 is further used for:

[0327] The operation data of the first operation is sent to the operation module, and the operation module is used to synchronize the operation data to the anti-fraud platform.

[0328] Optionally, the processor 710 is further configured to:

[0329] Obtaining the anti-fraud operation data of the anti-fraud platform for the call;

[0330] The LLM model is iteratively trained according to the operation data of the first operation and the anti-fraud operation data.

[0331] Optionally, the transceiver 720 is further used for:

[0332] Receive the fourth request message sent by the AS module, where the fourth request message is used to request the agent session processing service.

[0333] It should be noted that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment applied to the core service module, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0334] like Figure 8 As shown, a communication device in an embodiment of the present application, the communication device may be a second communication device 800, the second communication device may be an AS module, including a processor 810 and a transceiver 820;

[0335] The transceiver 820 is used to: when simulating a virtual holder of the first number to talk with the first terminal, send a first request message to the core service module according to the configuration information of the virtual holder, wherein the first request message is used to request response information from the core service module;

[0336] receiving a response message sent by the core service module, wherein the response message is related to the configuration information of the virtual holder and the call context information;

[0337] Sending the response information to the first terminal.

[0338] Optionally, the transceiver 820 is further used for:

[0339] Sending a second request message to the media capability module, where the second request message carries voice call context information of a call between the AS module and the first terminal, and the second request message is used to request the media capability module to convert the voice call context information into text call context information;

[0340] Receive text call context information sent by the media capability module.

[0341] Optionally, the transceiver 820 is further used for:

[0342] Sending a third request message to the media capability module, the third request message carrying text response information, the third request message being used to request the media capability module to convert the text response information into voice response information;

[0343] The sending the response information to the first terminal includes:

[0344] The voice response information is sent to the first terminal.

[0345] Optionally, the first request message is further used to request the core service module to perform an anti-fraud related operation, where the anti-fraud related operation includes at least one of the following:

[0346] Fraud detection through semantic analysis;

[0347] Conduct data collection;

[0348] The first terminal is numbered.

[0349] Optionally, the transceiver 820 is further used for:

[0350] When the call with the first terminal is connected, if the call meets the first condition, first indication information is sent to the core processing module, where the first indication information is used to instruct the core processing module to perform audio and video anchoring and media negotiation with the first terminal and the AS module;

[0351] The first condition includes at least one of the following:

[0352] The call is an audio or video call;

[0353] The terminal where the AS module is located supports audio and video calls;

[0354] The terminal where the AS module is located supports call service preprocessing.

[0355] Optionally, the transceiver 820 is further used for:

[0356] A fourth request message is sent to the core service module, where the fourth request message is used to request the agent session processing service.

[0357] Optionally, the transceiver 820 is further used for:

[0358] The configuration information of the virtual holder of the first number is obtained from the operation module, wherein the configuration information of the virtual holder of each number in the number set is stored in the operation module.

[0359] It should be noted that the above-mentioned device provided in the embodiment of the present application can implement all the method steps implemented in the above-mentioned method embodiment applied to the AS module, and can achieve the same technical effect. The parts and beneficial effects of this embodiment that are the same as the method embodiment will not be described in detail here.

[0360] Another embodiment of the present invention further provides a communication device, which may be a core service module or an AS module. Fig. 9 As shown, it includes a transceiver 910, a processor 900, a memory 920, and a program or instruction stored in the memory 920 and executable on the processor 900; when the processor 900 executes the program or instruction, the above-mentioned anti-fraud call method is implemented.

[0361] The transceiver 910 is used to receive and send data under the control of the processor 900 .

[0362] Among them, Fig. 9In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically linking together various circuits of one or more processors represented by processor 900 and memory represented by memory 920. The bus architecture may also link together various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 910 may be a plurality of components, i.e., including a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium. The processor 900 is responsible for managing the bus architecture and general processing, and the memory 920 may store data used by the processor 900 when performing operations.

[0363] An embodiment of the present application also provides a readable storage medium on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps in the anti-fraud call method as described above are implemented and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0364] The processor is a processor in the network device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0365] An embodiment of the present application also provides a computer program product on which a program or instruction is stored, including computer instructions. When the computer instructions are executed by a processor, the steps in the anti-fraud call method as described above are implemented and the same technical effect can be achieved. To avoid repetition, they will not be repeated here.

[0366] It should be further explained that the terminal devices described in this specification include but are not limited to smart phones, tablet computers, etc., and many of the functional components described are called modules in order to more particularly emphasize the independence of their implementation methods.

[0367] In the embodiment of the present application, module can be implemented with software so that it can be executed by various types of processors. For example, an executable code module of an identification can include one or more physical or logical blocks of computer instructions, for example, it can be constructed as an object, process or function. Nevertheless, the executable code of the identified module does not need to be physically located together, but can include different instructions stored in different positions, and when these instructions are logically combined together, it constitutes a module and realizes the specified purpose of the module.

[0368] In fact, executable code module can be a single instruction or many instructions, and can even be distributed on a plurality of different code segments, distributed among different programs, and distributed across a plurality of memory devices. Similarly, operating data can be identified in the module, and can be implemented and organized in the data structure of any appropriate type according to any appropriate form. The operating data can be collected as a single data set, or can be distributed in different locations (including on different storage devices), and can only be present on a system or network as an electronic signal at least in part.

[0369] When a module can be implemented by software, considering the level of existing hardware technology, a person skilled in the art can build a corresponding hardware circuit to implement the corresponding function of the module that can be implemented by software without considering the cost. The hardware circuit includes a conventional very large scale integration (VLSI) circuit or gate array and existing semiconductors such as logic chips, transistors, or other discrete components. The module can also be implemented by a programmable hardware device, such as a field programmable gate array, a programmable array logic, a programmable logic device, etc.

[0370] The above exemplary embodiments are described with reference to the accompanying drawings, and many different forms and embodiments are feasible without departing from the spirit and teachings of the present application. Therefore, the present application should not be constructed as a limitation of the exemplary embodiments proposed herein. More specifically, these exemplary embodiments are provided so that the present application will be perfect and complete, and the scope of the present application will be conveyed to those who are familiar with the technology. In these figures, the component sizes and relative sizes may be exaggerated for clarity. The terms used here are only based on the purpose of describing specific exemplary embodiments and are not intended to be limiting. As used herein, unless the text clearly indicates otherwise, the singular forms "one", "an" and "the" are intended to include these multiple forms. It will be further understood that the terms "comprising" and / or "including" when used in this specification indicate the presence of the features, integers, steps, operations, components and / or components, but do not exclude the presence or increase of one or more other features, integers, steps, operations, components, components and / or their groups. Unless otherwise indicated, when stated, a range of values ​​includes the upper and lower limits of that range and any subranges therebetween.

[0371] The above is a preferred embodiment of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles described in the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for preventing fraudulent calls. It is characterized in that Applied to core service modules, including: The core service module receives a first request message sent by the application service AS module, wherein the first request message is used to request: response information of the AS module when simulating a virtual holder of the first number to talk with the first terminal; Perform a first operation according to the call context information and the configuration information of the virtual holder; The first operation includes at least one of the following: Generate response information; Fraud detection through semantic analysis; Conduct data collection; The first terminal is numbered.

2. The method according to claim 1, It is characterized in that In the case where the first operation includes generating a response message, performing the first operation according to the call context information and the configuration information of the virtual holder includes: generating a first prompt word according to the call context information and the configuration information of the virtual holder, wherein the first prompt word is a keyword that matches the virtual holder and is related to the response information; According to the call context information, a search is performed in a database using a search enhancement generation RAG module to obtain search information, where the search information is a response content related to the call context information in the database; generating a second prompt word according to the call context information, the first prompt word and the search information; The second prompt word is input into the large language model LLM to obtain response information.

3. The method according to claim 1, It is characterized in that The method further comprises: The operation data of the first operation is sent to the operation module, and the operation module is used to synchronize the operation data to the anti-fraud platform.

4. The method according to any one of claims 1 to 3, It is characterized in that The method further comprises: Obtaining the anti-fraud operation data of the anti-fraud platform for the call; The LLM model is iteratively trained according to the operation data of the first operation and the anti-fraud operation data.

5. The method according to claim 1, It is characterized in that The method further comprises: Receive the fourth request message sent by the AS module, where the fourth request message is used to request the agent session processing service. 6.An anti-fraud call method, It is characterized in that Applicable to the application service AS module, including: When simulating a virtual holder of the first number to talk to the first terminal, the AS module sends a first request message to the core service module according to the configuration information of the virtual holder, where the first request message is used to request response information from the core service module; receiving a response message sent by the core service module, wherein the response message is related to the configuration information of the virtual holder and the call context information; Sending the response information to the first terminal.

7. The method according to claim 6, It is characterized in that Before sending the first request message to the core service module, the method further includes: Sending a second request message to the media capability module, where the second request message carries voice call context information of a call between the AS module and the first terminal, and the second request message is used to request the media capability module to convert the voice call context information into text call context information; Receive text call context information sent by the media capability module.

8. The method according to claim 6, It is characterized in that Before sending the response information to the first terminal, the method further includes: Sending a third request message to the media capability module, the third request message carrying text response information, the third request message being used to request the media capability module to convert the text response information into voice response information; The sending the response information to the first terminal includes: The voice response information is sent to the first terminal.

9. The method according to claim 6, It is characterized in that The first request message is also used to request the core service module to perform an anti-fraud related operation, and the anti-fraud related operation includes at least one of the following: Fraud detection through semantic analysis; Conduct data collection; The first terminal is numbered.

10. The method according to claim 6, It is characterized in that The method further comprises: When the call with the first terminal is connected, if the call meets the first condition, first indication information is sent to the core processing module, where the first indication information is used to instruct the core processing module to perform audio and video anchoring and media negotiation with the first terminal and the AS module; The first condition includes at least one of the following: The call is an audio or video call; The terminal where the AS module is located supports audio and video calls; The terminal where the AS module is located supports call service preprocessing.

11. The method according to claim 6, It is characterized in that Before sending the first request message to the core service module, the method further includes: A fourth request message is sent to the core service module, where the fourth request message is used to request the agent session processing service.

12. The method according to claim 6, It is characterized in that Before sending the first request message to the core service module, the method further includes: The configuration information of the virtual holder of the first number is obtained from the operation module, wherein the configuration information of the virtual holder of each number in the number set is stored in the operation module. 13.An anti-fraud system, It is characterized in that include: The AS module is used to send a first request message to the core service module according to the configuration information of the virtual holder when simulating a virtual holder of the first number to talk with the first terminal, wherein the first request message is used to request response information from the core service module; The core service module is used to receive the first request message sent by the AS module; and perform a first operation according to the call context information and the configuration information of the virtual holder; The first operation includes at least one of the following: Generate response information; Fraud detection through semantic analysis; Conduct data collection; Marking the first terminal with a number; The AS module is also used for: receiving response information sent by the core service module; and sending the response information to the first terminal.

14. The system according to claim 13, It is characterized in that The system further comprises: The media capability module is used to convert the voice call context information sent by the AS module into text call context information; and / or to convert the text response information sent by the AS module into voice response information.

15. The system according to claim 13, It is characterized in that The system further comprises: The operation module is used to synchronize the operation data of the first operation to the anti-fraud platform. 16.An anti-fraud call device, It is characterized in that include: The core service module is used to receive a first request message sent by the application service AS module, wherein the first request message is used to request: response information of the AS module when simulating a virtual holder of the first number to talk with the first terminal; Perform a first operation according to the call context information and the configuration information of the virtual holder; The first operation includes at least one of the following: Generate response information; Fraud detection through semantic analysis; Conduct data collection; The first terminal is numbered. 17.An anti-fraud call device, It is characterized in that include: The AS module is used to send a first request message to the core service module according to the configuration information of the virtual holder when simulating a virtual holder of the first number to talk with the first terminal, wherein the first request message is used to request response information from the core service module; receiving a response message sent by the core service module, wherein the response message is related to the configuration information of the virtual holder and the call context information; Sending the response information to the first terminal.

18. A communication device, It is characterized in that It includes a transceiver, a processor, a memory, and a program or instruction stored in the memory and executable on the processor; when the processor executes the program or instruction, the steps of the anti-fraud call method as described in any one of claims 1 to 5 are implemented, or the steps of the anti-fraud call method as described in any one of claims 6 to 12 are implemented.

19. A readable storage medium having a program or instruction stored thereon, It is characterized in that When the program or instruction is executed by the processor, the steps of the anti-fraud call method as described in any one of claims 1 to 5 are implemented, or the steps of the anti-fraud call method as described in any one of claims 6 to 12 are implemented.

20. A computer program product having a program or instructions stored thereon, It is characterized in that The method comprises computer instructions, which, when executed by a processor, implement the steps of the anti-fraud call method as described in any one of claims 1 to 5, or implement the steps of the anti-fraud call method as described in any one of claims 6 to 12.

Images