Risk assessment method, device and equipment based on large model Agent and storage medium

Abstract

The invention relates to the technical field of natural language processing and financial science and technology, and particularly discloses a risk assessment method and device based on a large model Agent, computer equipment and a storage medium. Obtaining a risk assessment task and a task demand corresponding to the to-be-assessed customer; and based on the large model agent Agent, decomposing the risk assessment task according to task requirements to obtain risk identification sub-tasks, carrying out risk identification according to a risk identification sequence to obtain a risk identification result of each sub-task, and combining a preset risk assessment rule to obtain a risk assessment result. According to the risk assessment method and device, the risk assessment task is decomposed through the large model agent to obtain the sub-tasks, and the sub-tasks are processed, so that automatic processing is realized, manual intervention and human errors are reduced, and the risk assessment efficiency and the result accuracy are improved. Secondly, the risk identification subtasks are processed in sequence, it is ensured that all the subtasks are executed in a reasonable sequence, and the efficiency and accuracy of risk assessment are improved.

Description

Technical Field

[0001] The present application relates to the fields of natural language processing and financial technology, and in particular to a risk assessment method, device, computer equipment and storage medium based on a large model agent. Background Art

[0002] In the financial field, risk assessment is a key link in ensuring transaction security, preventing fraud and maintaining market order. However, traditional risk assessment methods require manual intervention and have single technical tools, which cannot meet the diverse risk assessment needs. Due to the huge amount of data and diverse types in the financial industry, and the limited data processing capabilities of business personnel, the probability of errors increases, which reduces the accuracy of risk assessment results. In addition, the query and analysis tasks in traditional risk assessment methods are cumbersome and time-consuming, which reduces the efficiency of risk assessment. Therefore, how to improve the efficiency of risk assessment and the accuracy of assessment results has become an urgent problem to be solved. Summary of the invention

[0003] The present application provides a risk assessment method, apparatus, computer equipment and storage medium based on a large model Agent to improve the efficiency of risk assessment and the accuracy of assessment results.

[0004] In a first aspect, the present application provides a risk assessment method based on a large model agent, the method comprising:

[0005] Obtaining the risk assessment tasks corresponding to the customers to be assessed and the task requirements of the risk assessment tasks;

[0006] Based on the large model intelligent agent Agent, the risk assessment task is decomposed according to the task requirements to obtain at least one risk identification subtask;

[0007] Obtaining a risk identification sequence for each of the risk identification subtasks, performing risk identification on each of the risk identification subtasks according to the risk identification sequence, and obtaining a risk identification result for each risk identification subtask;

[0008] Based on the risk identification results of each of the risk identification subtasks and preset risk assessment rules, a risk assessment result of the customer to be assessed is obtained.

[0009] In a second aspect, the present application also provides a risk assessment device based on a large model Agent, the device comprising:

[0010] A task requirement acquisition module is used to acquire the risk assessment task corresponding to the customer to be assessed and the task requirement of the risk assessment task;

[0011] An assessment task decomposition module is used to decompose the risk assessment task according to the task requirements based on the large model intelligent agent Agent to obtain at least one risk identification subtask;

[0012] An identification result obtaining module is used to obtain a risk identification sequence for each of the risk identification subtasks, perform risk identification on each of the risk identification subtasks according to the risk identification sequence, and obtain a risk identification result for each risk identification subtask;

[0013] The assessment result obtaining module is used to obtain the risk assessment result of the customer to be assessed based on the risk identification result of each risk identification subtask and the preset risk assessment rules.

[0014] In a third aspect, the present application also provides a computer device, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the risk assessment method based on a large model agent as described above when executing the computer program.

[0015] In a fourth aspect, the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the risk assessment method based on a large model agent as described above.

[0016] The present application discloses a risk assessment method, device, computer equipment and storage medium based on a large model agent, which obtains the risk assessment task corresponding to the customer to be assessed and the task requirements of the risk assessment task; based on the large model intelligent agent, the risk assessment task is decomposed according to the task requirements to obtain at least one risk identification subtask; the risk identification order of each risk identification subtask is obtained, and the risk identification of each risk identification subtask is performed according to the risk identification order to obtain the risk identification result of each risk identification subtask; based on the risk identification result of each risk identification subtask and the preset risk assessment rules, the risk assessment result of the customer to be assessed is obtained. The present application decomposes the risk assessment task to obtain subtasks through a large model intelligent agent agent, and processes the subtasks, thereby realizing automated processing, reducing manual intervention, reducing human errors, and thus improving the efficiency of risk assessment and the accuracy of the assessment results. Secondly, the risk identification subtasks are processed in sequence to ensure that the risk identification subtasks are executed in a reasonable order, thereby improving the efficiency and accuracy of risk assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 It is a schematic flow chart of a first embodiment of a risk assessment method based on a large model Agent provided in an embodiment of the present application;

[0019] Figure 2 is a schematic flow chart of a second embodiment of a risk assessment method based on a large model Agent provided in an embodiment of the present application;

[0020] Figure 3 is a schematic flow chart of a third embodiment of a risk assessment method based on a large model Agent provided in an embodiment of the present application;

[0021] Figure 4 A schematic block diagram of a risk assessment device based on a large model Agent provided in an embodiment of the present application;

[0022] Figure 5 A schematic block diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0024] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0025] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0026] It should be further understood that the term “and / or” used in the specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0027] The embodiments of the present application provide a risk assessment method, device, computer equipment and storage medium based on a large model agent. Among them, the risk assessment method based on a large model agent can be applied to a server, and the risk assessment task is decomposed by the large model intelligent agent to obtain subtasks, and the subtasks are processed, thereby realizing automated processing, reducing manual intervention, reducing human errors, and thus improving the efficiency of risk assessment and the accuracy of the assessment results. Secondly, the risk identification subtasks are processed in sequence to ensure that the risk identification subtasks are executed in a reasonable order, thereby improving the efficiency and accuracy of risk assessment. Among them, the server can be an independent server or a server cluster.

[0028] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0029] See also Figure 1 , Figure 1 This is a schematic flow chart of a risk assessment method based on a large model agent provided in an embodiment of the present application. The risk assessment method based on a large model agent can be applied to a server, and is used to decompose the risk assessment task through the large model intelligent agent to obtain subtasks, and process the subtasks, thereby realizing automated processing, reducing manual intervention, reducing human errors, and thus improving the efficiency of risk assessment and the accuracy of the assessment results. Secondly, the risk identification subtasks are processed in sequence to ensure that the risk identification subtasks are executed in a reasonable order, thereby improving the efficiency and accuracy of risk assessment.

[0030] like Figure 1 As shown, the risk assessment method based on the large model Agent specifically includes steps S101 to S104.

[0031] S101, obtaining the risk assessment task corresponding to the customer to be assessed and the task requirements of the risk assessment task;

[0032] In one embodiment, the customer to be assessed is a customer who needs to undergo a risk assessment.

[0033] In one embodiment, risk assessment tasks and task requirements can be obtained by analyzing the natural description language content input by the user. Specifically, the meaning of the natural description language content, including context, implicit meaning and user's goals, is understood by the large model Agent to determine the user's intention. Key information is extracted from the user's intention and converted into specific task requirements, such as "assessing the market risk of customer A" or "identifying the credit risk of customer B". Among them, the large model Agent is an intelligent agent based on a large language model. It is trained through deep learning technology and has strong natural language understanding and reasoning capabilities. It can understand and generate natural language.

[0034] In one embodiment, according to task requirements, speech to be recognized that is relevant to risk assessment is extracted from audio content corresponding to the customer to be recognized, and a risk assessment task is generated by combining the task requirements and the speech to be recognized.

[0035] S102, based on the large model intelligent agent Agent, decompose the risk assessment task according to the task requirements to obtain at least one risk identification subtask;

[0036] In a specific embodiment, the large model Agent parses the task requirements, identifies and defines the key subtasks of risk assessment, and decomposes the risk assessment task into multiple risk identification subtasks, such as voiceprint verification subtasks, speech recognition subtasks, etc.

[0037] S103, obtaining a risk identification sequence for each of the risk identification subtasks, performing risk identification on each of the risk identification subtasks according to the risk identification sequence, and obtaining a risk identification result for each risk identification subtask;

[0038] Furthermore, before obtaining the risk identification order of each risk identification subtask and performing risk identification on each risk identification subtask according to the risk identification order, and obtaining the risk identification result of each risk identification subtask, it also includes: obtaining the task logic and data dependency of each risk identification subtask; and sorting the risk identification subtasks based on the task logic and the data dependency to obtain the risk identification order of the risk identification subtasks.

[0039] In one embodiment, the logic of each risk identification subtask is analyzed to determine the task's objectives, required input data, processing steps, and expected outputs. Data dependencies between subtasks are determined, for example, the output of some subtasks may be the input of other subtasks.

[0040] In one embodiment, the risk identification subtasks are sorted according to the logical relationship and data dependency of each subtask, ensuring that the data it depends on is available before any subtask starts. Specifically, a topological sorting algorithm can be used to sort the subtasks.

[0041] In one embodiment, subtasks that have no logical relationship and data dependency with other subtasks can be processed in parallel to improve processing efficiency. They can also be sorted according to priority rules, where the priority rules are set by the user according to actual needs.

[0042] Furthermore, performing risk identification on each of the risk identification subtasks in the risk identification order to obtain a risk identification result for each risk identification subtask includes: obtaining a risk identification tool or a risk identification model corresponding to each of the risk identification subtasks; based on the large model Agent, calling the risk identification tool or the risk identification model in the risk identification order to process the speech to be recognized, and obtaining a risk identification result for each risk identification subtask.

[0043] In one embodiment, risk identification tools and risk identification models can be stored in a preset tool library, and risk identification tools and risk identification models can handle different types of risk assessment tasks. Match appropriate risk identification tools or models for each risk identification subtask. For example, the voiceprint recognition subtask corresponding to customer identity verification requires the use of a voiceprint recognition model to compare the speaker's voice characteristics, identify the customer's identity, and use a data query tool to query the database to see if the customer information is consistent with the information said by the customer in the voice to be identified. The subtask of assessing whether a customer has a fraud risk requires the use of a speech recognition model to identify whether there is relevant fraud speech in the customer's voice to be identified.

[0044] In a specific embodiment, different risk identification subtasks will perform different processing on the speech to be recognized. For example, the voiceprint verification subtask needs to pre-process the audio content such as noise reduction, enhancement, and denoising, and then use the voiceprint recognition model to recognize the voiceprint features of the speech to be recognized; the speech recognition subtask needs to convert the speech to be recognized into text, and use the speech recognition model to recognize the obtained text content to obtain the speech recognition result, match the recognized speech with the fraud speech, and determine the fraud risk score according to the degree of matching as the risk identification result.

[0045] In one embodiment, according to the risk identification order of the risk identification subtasks, corresponding risk identification tools or models are called in sequence to process the audio content or the converted text.

[0046] In one embodiment, after each tool or model processes the audio content, it analyzes its output and extracts risk factors, and records the risk identification results of each subtask, including risk description, risk level, potential impact, etc.

[0047] In the above embodiment, it can be ensured that the risk identification subtasks are executed in a reasonable order, thereby improving the efficiency and accuracy of risk assessment.

[0048] S104: Based on the risk identification results of each of the risk identification subtasks and preset risk assessment rules, obtain the risk assessment results of the customer to be assessed.

[0049] In one embodiment, the risk identification results of all risk identification subtasks are obtained, and the risk assessment rules are clarified, including the classification of risk levels, the setting of risk weights, the identification of risk signals, etc.

[0050] In one embodiment, according to the risk assessment rules, each risk identification result is analyzed to determine whether there is a high-risk subtask. If there is a high-risk subtask, the customer to be assessed is determined as a high-risk customer; if there is no high-risk subtask, the risk scores of different subtasks are weighted averaged according to the preset weight coefficient to obtain a comprehensive risk score. The calculated risk score is compared with the preset risk level threshold. According to the comparison result, the customer is divided into the corresponding risk level, such as low risk, medium risk, high risk, etc.

[0051] In one embodiment, the risk assessment results are fed back to the user so that the user can further review and make decisions based on the risk assessment results, and formulate corresponding risk management strategies and measures.

[0052] The above embodiment provides a risk assessment method, device, computer equipment and storage medium based on a large model agent, which obtains the risk assessment task corresponding to the customer to be assessed and the task requirements of the risk assessment task; based on the large model agent, the risk assessment task is decomposed according to the task requirements to obtain at least one risk identification subtask; the risk identification order of each risk identification subtask is obtained, and the risk identification of each risk identification subtask is performed according to the risk identification order to obtain the risk identification result of each risk identification subtask; based on the risk identification result of each risk identification subtask and the preset risk assessment rules, the risk assessment result of the customer to be assessed is obtained. This application decomposes the risk assessment task to obtain subtasks through a large model agent agent, and processes the subtasks, thereby realizing automated processing, reducing manual intervention, reducing human errors, and thus improving the efficiency of risk assessment and the accuracy of the assessment results. Secondly, the risk identification subtasks are processed in sequence to ensure that the risk identification subtasks are executed in a reasonable order, thereby improving the efficiency and accuracy of risk assessment.

[0053] See also Figure 2 , Figure 2This is a schematic flow chart of a risk assessment method based on a large model agent provided in an embodiment of the present application. The risk assessment method based on a large model agent can be applied to a server, and is used to analyze the user's natural language description content and the audio content of the customer to be assessed through a large model intelligent agent, obtain risk assessment tasks and task requirements of risk assessment tasks, and fully understand the user's risk assessment requirements, obtain accurate risk assessment tasks, and improve the accuracy of risk assessment results.

[0054] like Figure 2 As shown, the risk assessment method based on the large model Agent specifically includes steps S201 to S202.

[0055] S201, obtaining the user's natural language description content and the audio content of the customer to be evaluated;

[0056] In one embodiment, the user refers to the person who issues the risk assessment task. The customer to be assessed is the customer who needs to undergo risk assessment. For example, the business personnel of a financial institution conducts risk assessment on customers who intend to purchase products to eliminate possible fraud risks. The business personnel are the users here, and the customers who intend to purchase products are the customers to be assessed.

[0057] In one embodiment, the natural language description content may be text content input by the user regarding the risk assessment requirements of the customer to be assessed, or may be text content obtained by converting the voice content input by the user into text. For example, the natural language description content may be "Please help me analyze whether the speaker in this voice is customer A himself, and assess whether he has a fraud risk."

[0058] In one embodiment, the audio content of the customer to be evaluated may be a recording file of a conversation between the customer to be evaluated and the business personnel, or may be a recording of a conversation between the customer to be evaluated and the business personnel.

[0059] S202: Based on the large model Agent, analyze the natural language description content and the audio content to obtain a risk assessment task and task requirements of the risk assessment task.

[0060] Furthermore, based on the large model Agent, the natural language description content and the audio content are analyzed to obtain a risk assessment task and a task requirement of the risk assessment task, including: based on the large model Agent, semantically parsing the natural language description content to obtain the task requirement; based on the task requirement, extracting the audio content to obtain the speech to be recognized corresponding to the task requirement; based on the task requirement and the speech to be recognized, generating the risk assessment task.

[0061] In one embodiment, the natural language description content may be cleaned, including removing irrelevant characters, punctuation marks, stop words, etc.

[0062] In one embodiment, the large model agent identifies keywords and phrases in the text, such as customer name, risk type, relevant date, etc. Based on the keywords and phrases, the agent understands the meaning of the text, including the context, implicit meaning, and user goals, and determines the user's intent, such as whether to assess specific risks, identify potential problems, or predict future risks.

[0063] In one embodiment, key information is extracted from the user's intent, and the extracted key information is converted into specific task requirements, such as "assessing the market risk of customer A" or "identifying the credit risk of customer B."

[0064] In one embodiment, the audio file provided by the user can be subjected to noise reduction, enhancement and format conversion. According to the task requirements, the speech to be recognized related to the risk assessment is extracted from the audio content. In combination with the task requirements and the speech to be recognized, the specific content and objectives of the risk assessment task are defined.

[0065] In the above embodiment, the large model intelligent agent is used to analyze the user's natural language description content and the audio content of the customer to be evaluated to obtain the risk assessment task and the task requirements of the risk assessment task, so as to fully understand the user's risk assessment requirements, obtain accurate risk assessment tasks, and improve the accuracy of the risk assessment results.

[0066] See also Figure 3 , Figure 3 This is a schematic flow chart of a risk assessment method based on a large model agent provided in an embodiment of the present application. The risk assessment method based on a large model agent can be applied to a server to automatically perform risk assessment tasks through a large model intelligent agent and provide risk assessment results based on rules and data driving, thereby reducing manual intervention and human errors and improving the efficiency and accuracy of risk assessment.

[0067] like Figure 3 As shown, the risk assessment method based on the large model Agent specifically includes steps S301 to S302.

[0068] S301, analyzing the risk identification results of each of the risk identification subtasks based on the risk assessment rules to determine whether there is a high-risk subtask in the risk identification subtask;

[0069] S302: If the high-risk subtask exists in the risk identification subtask, taking the customer to be evaluated as a high-risk customer as the risk assessment result.

[0070] Furthermore, based on the risk assessment rules, after analyzing the risk identification results of each of the risk identification subtasks to determine whether there are high-risk subtasks in the risk identification subtasks, it also includes: if the risk identification subtasks are all non-high-risk subtasks, obtaining a preset weight coefficient; based on the weight coefficient, weighting the risk identification results of each of the risk identification subtasks to obtain a risk assessment score for the risk assessment task; when the risk assessment score is higher than or equal to a preset risk score threshold, treating the customer to be assessed as a high-risk customer as the risk assessment result.

[0071] In one embodiment, the risk assessment rules can be freely set by the user according to actual needs. The risk assessment rules are used to analyze the risk identification results of each risk identification subtask, determine whether it is a high-risk subtask, and comprehensively determine whether the customer to be assessed in the risk assessment task is a high-risk customer.

[0072] For example, a high-risk keyword can be set in the risk assessment rule. When the high-risk keyword exists in the identification result of the risk identification subtask, the subtask is determined as a high-risk subtask. It can also be stipulated that when the number of high-risk subtasks in the risk assessment task reaches a certain threshold, the customer to be assessed in the risk assessment task is determined as a high-risk customer.

[0073] In one embodiment, if there is a high-risk subtask, the customer to be evaluated is marked as a high-risk customer. If all subtasks are non-high-risk subtasks, a preset weight coefficient corresponding to each subtask is obtained for weighted processing to obtain a risk score.

[0074] In a specific embodiment, a preset weight coefficient is obtained, wherein the coefficient can be freely set by the user based on the importance of the subtask and the degree of impact on the overall risk. The risk identification results of each non-high-risk subtask are weighted and a weighted score is calculated. The weighted scores are added together to obtain the risk assessment score of the entire risk assessment task.

[0075] In one embodiment, the calculated risk assessment score is compared with a preset risk score threshold. If the risk assessment score is higher than or equal to the risk score threshold, the customer to be assessed is determined to be a high-risk customer. The risk score threshold can be freely set by the user according to actual needs.

[0076] In the above embodiments, the large model intelligent agent can automatically perform risk assessment tasks and provide rule-based and data-driven risk assessment results, thereby reducing manual intervention and human errors and improving the efficiency and accuracy of risk assessment.

[0077] See also Figure 4 , Figure 4 The embodiment of the present application provides a schematic block diagram of a risk assessment device based on a large model Agent, which is used to execute the aforementioned risk assessment method based on a large model Agent. The risk assessment device based on a large model Agent can be configured on a server.

[0078] like Figure 4 As shown, the risk assessment device 400 based on the large model Agent includes:

[0079] The task requirement acquisition module 401 is used to acquire the risk assessment task corresponding to the client to be assessed and the task requirement of the risk assessment task;

[0080] An assessment task decomposition module 402 is used to decompose the risk assessment task according to the task requirements based on the large model agent Agent to obtain at least one risk identification subtask;

[0081] The identification result obtaining module 403 is used to obtain the risk identification sequence of each risk identification subtask, perform risk identification on each risk identification subtask according to the risk identification sequence, and obtain the risk identification result of each risk identification subtask;

[0082] The assessment result obtaining module 404 is used to obtain the risk assessment result of the client to be assessed based on the risk identification result of each risk identification subtask and the preset risk assessment rules.

[0083] Furthermore, the task requirement acquisition module 401 includes:

[0084] A content acquisition unit, used to acquire the user's natural language description content and the audio content of the customer to be evaluated;

[0085] The content analysis unit is used to analyze the natural language description content and the audio content based on the large model Agent to obtain the risk assessment task and the task requirements of the risk assessment task.

[0086] Furthermore, the content analysis unit includes:

[0087] A semantic parsing subunit, configured to perform semantic parsing on the natural language description content based on the large model Agent to obtain the task requirements;

[0088] An audio extraction subunit, configured to extract the audio content based on the task requirements to obtain the speech to be recognized corresponding to the task requirements;

[0089] The task generation subunit is used to generate the risk assessment task based on the task requirement and the speech to be recognized.

[0090] Furthermore, the identification result obtaining module 403 includes:

[0091] A model acquisition unit, used to acquire a risk identification tool or a risk identification model corresponding to each of the risk identification subtasks;

[0092] The content processing unit is used to call the risk identification tool or the risk identification model to process the speech to be identified based on the large model Agent in accordance with the risk identification sequence to obtain the risk identification result of each risk identification subtask.

[0093] Furthermore, the evaluation result obtaining module 404 includes:

[0094] A result analysis unit, configured to analyze the risk identification result of each of the risk identification subtasks based on the risk assessment rule, and determine whether there is a high-risk subtask in the risk identification subtask;

[0095] A result determination unit is configured to determine, if the high-risk subtask exists in the risk identification subtask, whether the customer to be evaluated is a high-risk customer as the risk assessment result.

[0096] Furthermore, the evaluation result obtaining module 404 further includes:

[0097] A weight coefficient acquisition unit, used for acquiring a preset weight coefficient if all of the risk identification subtasks are non-high-risk subtasks;

[0098] A weighted processing unit, configured to perform weighted processing on the risk identification result of each of the risk identification subtasks based on the weight coefficient to obtain a risk assessment score for the risk assessment task;

[0099] The result determination unit is used to determine, when the risk assessment score is higher than or equal to a preset risk score threshold, that the customer to be assessed is a high-risk customer as the risk assessment result.

[0100] Furthermore, the risk assessment device 400 based on the large model agent further includes:

[0101] A task logic acquisition unit, used to acquire the task logic and data dependency of each of the risk identification subtasks;

[0102] The risk identification sequence obtaining unit is used to sort the risk identification subtasks based on the task logic and the data dependency to obtain the risk identification sequence of the risk identification subtasks.

[0103] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0104] The above-mentioned device can be implemented in the form of a computer program. Figure 5 Runs on the computer device shown.

[0105] See also Figure 5 , Figure 5 1 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device may be a server.

[0106] See also Figure 5 The computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.

[0107] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any risk assessment method based on a large model agent.

[0108] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0109] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any risk assessment method based on the large model Agent.

[0110] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0111] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0112] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0113] Obtaining the risk assessment tasks corresponding to the customers to be assessed and the task requirements of the risk assessment tasks;

[0114] Based on the large model intelligent agent Agent, the risk assessment task is decomposed according to the task requirements to obtain at least one risk identification subtask;

[0115] Obtaining a risk identification sequence for each of the risk identification subtasks, performing risk identification on each of the risk identification subtasks according to the risk identification sequence, and obtaining a risk identification result for each risk identification subtask;

[0116] Based on the risk identification results of each of the risk identification subtasks and preset risk assessment rules, a risk assessment result of the customer to be assessed is obtained.

[0117] In one embodiment, when the processor implements obtaining the risk assessment task corresponding to the customer to be assessed and the task requirements of the risk assessment task, it is used to implement:

[0118] Obtaining the user's natural language description content and the audio content of the customer to be evaluated;

[0119] Based on the large model Agent, the natural language description content and the audio content are analyzed to obtain a risk assessment task and task requirements of the risk assessment task.

[0120] In one embodiment, when the processor analyzes the natural language description content and the audio content based on the large model Agent to obtain the risk assessment task and the task requirements of the risk assessment task, it is used to implement:

[0121] Based on the large model Agent, semantically analyze the natural language description content to obtain the task requirements;

[0122] Based on the task requirements, the audio content is extracted to obtain the speech to be recognized corresponding to the task requirements;

[0123] The risk assessment task is generated based on the task requirement and the speech to be recognized.

[0124] In one embodiment, when the processor performs risk identification on each of the risk identification subtasks according to the risk identification order and obtains the risk identification result of each risk identification subtask, it is used to implement:

[0125] Obtaining a risk identification tool or a risk identification model corresponding to each of the risk identification subtasks;

[0126] Based on the large model Agent, the risk identification tool or the risk identification model is called according to the risk identification sequence to process the speech to be identified, so as to obtain the risk identification result of each risk identification subtask.

[0127] In one embodiment, when the processor obtains the risk assessment result of the customer to be assessed based on the risk identification results of each risk identification subtask and the preset risk assessment rules, it is used to implement:

[0128] Based on the risk assessment rules, analyzing the risk identification results of each of the risk identification subtasks to determine whether there is a high-risk subtask in the risk identification subtask;

[0129] If the high-risk subtask exists in the risk identification subtask, the customer to be evaluated is regarded as a high-risk customer as the risk assessment result.

[0130] In one embodiment, after analyzing the risk identification result of each risk identification subtask based on the risk assessment rule to determine whether there is a high-risk subtask in the risk identification subtask, the processor is further configured to implement:

[0131] If the risk identification subtasks are all non-high-risk subtasks, then a preset weight coefficient is obtained;

[0132] Based on the weight coefficient, weighted processing is performed on the risk identification result of each of the risk identification subtasks to obtain a risk assessment score for the risk assessment task;

[0133] When the risk assessment score is higher than or equal to a preset risk score threshold, the customer to be assessed is regarded as a high-risk customer as the risk assessment result.

[0134] In one embodiment, before obtaining the risk identification order of each risk identification subtask, performing risk identification on each risk identification subtask according to the risk identification order, and obtaining the risk identification result of each risk identification subtask, the processor is further configured to implement:

[0135] Obtaining the task logic and data dependency of each of the risk identification subtasks;

[0136] The risk identification subtasks are sorted based on the task logic and the data dependency to obtain a risk identification sequence of the risk identification subtasks.

[0137] A computer-readable storage medium is also provided in an embodiment of the present application, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and the processor executes the program instructions to implement any of the large model agent-based risk assessment methods provided in the embodiments of the present application.

[0138] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc., equipped on the computer device.

[0139] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A risk assessment method based on a large model agent, characterized in that: include: Obtaining the risk assessment tasks corresponding to the customers to be assessed and the task requirements of the risk assessment tasks; Based on the large model intelligent agent Agent, the risk assessment task is decomposed according to the task requirements to obtain at least one risk identification subtask; Obtaining a risk identification sequence for each of the risk identification subtasks, performing risk identification on each of the risk identification subtasks according to the risk identification sequence, and obtaining a risk identification result for each risk identification subtask; Based on the risk identification results of each of the risk identification subtasks and preset risk assessment rules, a risk assessment result of the customer to be assessed is obtained.

2. The risk assessment method based on large model Agent according to claim 1 is characterized in that: The step of obtaining the risk assessment task corresponding to the client to be assessed and the task requirements of the risk assessment task includes: Obtaining the user's natural language description content and the audio content of the customer to be evaluated; Based on the large model Agent, the natural language description content and the audio content are analyzed to obtain a risk assessment task and task requirements of the risk assessment task.

3. The risk assessment method based on large model Agent according to claim 2 is characterized in that: The large model Agent is based on which the natural language description content and the audio content are analyzed to obtain a risk assessment task and the task requirements of the risk assessment task, including: Based on the large model Agent, semantically analyze the natural language description content to obtain the task requirements; Based on the task requirements, the audio content is extracted to obtain the speech to be recognized corresponding to the task requirements; The risk assessment task is generated based on the task requirement and the speech to be recognized.

4. The risk assessment method based on large model Agent according to claim 2 is characterized in that: The step of performing risk identification on each of the risk identification subtasks in the risk identification sequence to obtain a risk identification result for each risk identification subtask includes: Obtaining a risk identification tool or a risk identification model corresponding to each of the risk identification subtasks; Based on the large model Agent, the risk identification tool or the risk identification model is called according to the risk identification sequence to process the speech to be identified, so as to obtain the risk identification result of each risk identification subtask.

5. The risk assessment method based on large model Agent according to claim 1 is characterized in that: The step of obtaining the risk assessment result of the customer to be assessed based on the risk identification result of each risk identification subtask and the preset risk assessment rules includes: Based on the risk assessment rules, analyzing the risk identification results of each of the risk identification subtasks to determine whether there is a high-risk subtask in the risk identification subtask; If the high-risk subtask exists in the risk identification subtask, the customer to be evaluated is regarded as a high-risk customer as the risk assessment result.

6. The risk assessment method based on large model Agent according to claim 5 is characterized in that: After analyzing the risk identification result of each risk identification subtask based on the risk assessment rule to determine whether there is a high-risk subtask in the risk identification subtask, the method further includes: If the risk identification subtasks are all non-high-risk subtasks, then a preset weight coefficient is obtained; Based on the weight coefficient, weighted processing is performed on the risk identification result of each of the risk identification subtasks to obtain a risk assessment score for the risk assessment task; When the risk assessment score is higher than or equal to a preset risk score threshold, the customer to be assessed is regarded as a high-risk customer as the risk assessment result.

7. The risk assessment method based on large model Agent according to any one of claims 1 to 6, characterized in that: Before obtaining the risk identification order of each risk identification subtask, performing risk identification on each risk identification subtask according to the risk identification order, and obtaining the risk identification result of each risk identification subtask, the method further includes: Obtaining the task logic and data dependency of each of the risk identification subtasks; The risk identification subtasks are sorted based on the task logic and the data dependency to obtain a risk identification sequence of the risk identification subtasks.

8. A risk assessment device based on a large model agent, characterized in that: include: A task requirement acquisition module is used to acquire the risk assessment task corresponding to the customer to be assessed and the task requirement of the risk assessment task; An assessment task decomposition module is used to decompose the risk assessment task according to the task requirements based on the large model intelligent agent Agent to obtain at least one risk identification subtask; An identification result obtaining module is used to obtain a risk identification sequence for each of the risk identification subtasks, perform risk identification on each of the risk identification subtasks according to the risk identification sequence, and obtain a risk identification result for each risk identification subtask; The assessment result obtaining module is used to obtain the risk assessment result of the customer to be assessed based on the risk identification result of each risk identification subtask and the preset risk assessment rules.

9. A computer device, characterized in that: The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is used to execute the computer program and implement the large model agent-based risk assessment method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor implements the risk assessment method based on a large model agent according to any one of claims 1 to 7.

Images