A method, apparatus, device and medium for training a collection personnel
By acquiring virtual characters to simulate debt collection dialogues and using large language models to analyze and generate reference scripts, the problem of poor adaptability of traditional dialogue templates is solved, thereby improving the success rate and efficiency of debt collection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN XIAOYUDIAN DIGITAL TECH CO LTD
- Filing Date
- 2026-01-15
- Publication Date
- 2026-05-29
AI Technical Summary
In existing debt collection operations, traditional dialogue templates are difficult to adapt to diverse debt collection scenarios, resulting in a low success rate in debt collection.
By acquiring virtual avatars of target debt collectors, conducting simulated debt collection dialogues with them, analyzing dialogue information using a pre-set large language model, generating reference scripts to guide debt collectors in conducting debt collection dialogues with target debt collectors, and optimizing dialogue quality through scoring and summary mechanisms.
It improves collection success rates in diverse collection scenarios, avoids reliance on traditional dialogue templates, and enhances the targeting and efficiency of collection personnel.
Smart Images
Figure CN122115093A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of lending, and more particularly to a method, apparatus, equipment, and medium for training debt collectors. Background Technology
[0002] With the development of the national economy, a wide variety of credit products have emerged in the market. These credit products provide users with convenient consumer finance services. However, as the scale of consumer credit continues to expand, the non-performing loan rate is also rising in tandem. More and more users are unable to repay their debts on time, giving rise to debt collection services.
[0003] Currently, the workflow for debt collection is relatively simple. First, a dialogue template is created based on historical collection experience. This template guides collectors in communicating with overdue customers. Then, collectors use the template to communicate with the overdue customers, thereby achieving the goal of recovering the debt.
[0004] However, different users have different reasons for overdue payments, personality traits, and income levels, making the aforementioned dialogue template difficult to adapt to diverse collection scenarios. In this situation, collection personnel struggle to conduct effective collection efforts targeting users in specific scenarios, thus impacting collection success rates. Summary of the Invention
[0005] This application provides a method, apparatus, equipment, and medium for training debt collectors, aiming to solve the technical problem that dialogue templates are difficult to adapt to diverse debt collection scenarios, resulting in a low debt collection success rate.
[0006] In a first aspect, embodiments of this application provide a method for training debt collectors, comprising: Obtain the virtual avatar of the target debt collector; Engage in a simulated debt collection dialogue with the virtual character to obtain dialogue information; The dialogue information and the first preset prompt word are input into the preset large language model to obtain the reference script corresponding to the debt collection simulation dialogue. The reference script is used to guide the debt collector to conduct a debt collection dialogue with the target debt collector. The first preset prompt word is used to instruct the preset large language model to analyze the reference script from the dialogue information. The preset large language model is used to analyze the reference script from the dialogue information based on the first preset prompt word.
[0007] Optionally, after engaging in a simulated debt collection dialogue with the virtual character and obtaining dialogue information, the method further includes: The dialogue information and the second preset prompt word are input into the preset large language model to obtain the score corresponding to the debt collection simulation dialogue. The score is used to characterize the quality level of the debt collection simulation dialogue between the debt collector and the virtual character. The second preset prompt word is used to instruct the preset large language model to score the debt collection simulation dialogue. The preset large language model is also used to score the debt collection simulation dialogue based on the second preset prompt word.
[0008] Optionally, after engaging in a simulated debt collection dialogue with the virtual character and obtaining dialogue information, the method further includes: The dialogue information and the third preset prompt word are input into the preset large language model to obtain the summary information corresponding to the debt collection simulation dialogue. The summary information is used to characterize the shortcomings of the debt collector in the debt collection simulation dialogue. The preset large language model is also used to summarize the dialogue information based on the third preset prompt word.
[0009] Optionally, obtaining the virtual persona of the target debt collector includes: Define the person template for the target debt collector; Based on the person template, the collection information of the target loan object is extracted from the information of the target loan object, and the target loan object corresponds to the target collection object; Based on the collection information, m collection cases that match the collection information are selected from a preset collection case library. The preset collection case library includes n collection cases, where n and m are positive integers and n is greater than or equal to m. The collection information and the information of the m collection cases are input into the preset large language model to obtain the virtual character of the target collection object.
[0010] Optionally, the person template includes personality, occupation, gender, age, relationship with the borrower, and loan scenario.
[0011] Optionally, the step of selecting m collection case information that match the collection information from a preset collection case database based on the collection information includes: Calculate the similarity between the collection information and each collection case information in the preset collection case database to obtain n similarity scores, and the n similarity scores correspond one-to-one with the n collection case information; Select the top m similarities from the n similarities in descending order of similarity. Obtain the collection case information corresponding to the first m similarities to obtain the m collection case information.
[0012] Optionally, the process of constructing the preset collection case library includes: Obtain audio recordings of conversations from multiple historical debt collection cases; The dialogue content is extracted from the voice recordings of each of the multiple historical collection cases to obtain multiple dialogue texts, and the multiple dialogue texts correspond one-to-one with the multiple historical collection cases. The multiple dialogue texts are input into the preset large language model to obtain multiple collection case information, and the multiple collection case information corresponds one-to-one with the multiple dialogue texts.
[0013] Secondly, embodiments of this application also provide a debt collection personnel training device, which includes a unit for performing the above-described method.
[0014] Thirdly, embodiments of this application also provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the above-described method.
[0015] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the above-described method.
[0016] This application provides a method, apparatus, device, and medium for training debt collectors. The method includes: acquiring a virtual character representing a target debtor; conducting a simulated debt collection dialogue with the virtual character to obtain dialogue information; inputting the dialogue information and a first preset prompt word into a preset large language model to obtain a reference script corresponding to the simulated debt collection dialogue. The reference script is used to guide debt collectors in conducting debt collection dialogues with the target debtor. The first preset prompt word instructs the preset large language model to analyze the dialogue information to obtain the reference script, and the preset large language model is used to analyze the dialogue information based on the first preset prompt word to obtain the reference script. Therefore, the technical solution of this application involves acquiring a virtual character representing a target debtor, conducting a simulated debt collection dialogue with the virtual character to obtain dialogue information, and finally inputting the dialogue information and the first preset prompt word into a preset large language model to obtain a reference script corresponding to the simulated debt collection dialogue. The first preset prompt word instructs the preset large language model to analyze the dialogue information to obtain the reference script, which guides debt collectors in conducting debt collection dialogues with the target debtor. Therefore, the virtual character in this application's technical solution can realistically simulate the target debtor. By engaging in simulated debt collection dialogues with them and analyzing the dialogue information using a pre-set large language model, the system can effectively extract the reference scripts needed for dialogues with the target debtor. This mechanism does not rely on traditional dialogue templates, can adapt to diverse debt collection scenarios, and thus effectively improves the debt collection success rate. Attached Figure Description
[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0019] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.
[0020] Figure 1 A flowchart illustrating a debt collection personnel training method provided in this application embodiment; Figure 2A schematic block diagram of a debt collector training device provided in this application embodiment; Figure 3 A computer device provided in an embodiment of this application. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0022] The following disclosure provides numerous different embodiments or examples for implementing various structures of this application. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the scope of this application. Furthermore, reference numerals and / or letters may be repeated in different examples. Such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed.
[0023] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.
[0024] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0025] It should also be further understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.
[0026] As used in this specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if [described condition or event] is detected" may be interpreted, depending on the context, as "once determined," "in response to determination," "once [described condition or event] is detected," or "in response to detection of [described condition or event]."
[0027] To address the technical problem that existing dialogue templates are difficult to adapt to diverse collection scenarios, resulting in low collection success rates, this application provides a collection personnel training device that can improve collection success rates.
[0028] Figure 1 This is a flowchart illustrating a debt collection personnel training method provided in an embodiment of this application. In one embodiment, the method includes steps S101-S103.
[0029] S101. Obtain the virtual character of the target debt collector.
[0030] The virtual character is a non-physical entity constructed based on digital technology, possessing specific identity attributes and interactive capabilities. This application's embodiment obtains a virtual character of the target debt collector. This virtual character can realistically simulate the behavior of the target debt collector.
[0031] S102. Engage in a simulated debt collection dialogue with a virtual character to obtain dialogue information.
[0032] In this embodiment, debt collectors can engage in one or more rounds of simulated debt collection dialogues with virtual avatars. Each round of simulated debt collection dialogue includes one or more questions. In this embodiment, debt collectors can engage in simulated debt collection dialogues with virtual avatars via voice or text, and record the content of each round of dialogue to obtain dialogue information for each round.
[0033] S103. Input the dialogue information and the first preset prompt into the preset large language model to obtain the reference script corresponding to the debt collection simulation dialogue.
[0034] The reference script is used to guide debt collectors in conducting collection conversations with target debtors. The first preset prompt word instructs the preset large language model to analyze the dialogue information and derive the reference script. The preset large language model is used to analyze the dialogue information based on the first preset prompt word to derive the reference script. The preset large language model can be a general-purpose large language model.
[0035] This application provides a method for training debt collectors. The method includes: acquiring a virtual character representing a target debtor; conducting a simulated debt collection dialogue with the virtual character to obtain dialogue information; inputting the dialogue information and a first preset prompt word into a preset large language model to obtain a reference script corresponding to the simulated debt collection dialogue. The reference script is used to guide debt collectors in conducting debt collection dialogues with the target debtor. The first preset prompt word instructs the preset large language model to analyze the dialogue information to obtain the reference script, and the preset large language model analyzes the dialogue information based on the first preset prompt word to obtain the reference script. Therefore, the technical solution of this application involves acquiring a virtual character representing a target debtor, conducting a simulated debt collection dialogue with the virtual character to obtain dialogue information, and finally inputting the dialogue information and the first preset prompt word into a preset large language model to obtain a reference script corresponding to the simulated debt collection dialogue. The first preset prompt word instructs the preset large language model to analyze the dialogue information to obtain the reference script, which guides debt collectors in conducting debt collection dialogues with the target debtor. Therefore, the virtual character in this application's technical solution can realistically simulate the target debtor. By engaging in simulated debt collection dialogues with them and analyzing the dialogue information using a pre-set large language model, the system can effectively extract the reference scripts needed for dialogues with the target debtor. This mechanism does not rely on traditional dialogue templates, can adapt to diverse debt collection scenarios, and thus effectively improves the debt collection success rate.
[0036] In one embodiment, after S102, the method further includes S104.
[0037] S104. Input the dialogue information and the second preset prompt into the preset large language model to obtain the score corresponding to the debt collection simulation dialogue.
[0038] The scoring system is used to characterize the quality level of the simulated debt collection dialogue between the debt collector and the virtual character. A second preset prompt word is used to instruct the preset large language model to score the simulated debt collection dialogue. In this embodiment, the simulated debt collection dialogue can be scored from different perspectives. For example, the second preset prompt word could be "Please score the debt collector from a compliance perspective," or it could be "Please score the debt collector from a debt collection skills perspective." Of course, in this embodiment, the simulated debt collection dialogue can also be scored from the perspective of the loan officer. This application does not impose any limitations. The preset large language model is also used to score the simulated debt collection dialogue based on the second preset prompt word.
[0039] This application embodiment scores simulated debt collection dialogues, enabling debt collectors to summarize these dialogues in a targeted manner and obtain better reference scripts.
[0040] In one embodiment, after S102, the method further includes: S105 S105 inputs the dialogue information and the third preset prompt word into the preset large language model to obtain the summary information corresponding to the debt collection simulation dialogue.
[0041] The summary information is used to characterize the shortcomings of debt collectors in simulated debt collection dialogues. The large language model is also used to summarize the dialogue information based on the third preset prompt words.
[0042] It should be noted that in this embodiment, collection personnel can not only summarize the simulated collection dialogue themselves, but also summarize it using a preset large language model, thereby improving the efficiency and accuracy of the summary. Collection personnel can use the summary information to identify shortcomings in the current collection dialogue, prompting them to improve their collection techniques and ultimately increase the collection success rate.
[0043] In one embodiment, S101 specifically includes the following steps: S1011-S1014.
[0044] S1011, Template of the person to be collected.
[0045] In one embodiment, the person template includes, but is not limited to, personality, occupation, gender, age, relationship with the borrower, and loan scenario. The loan scenario can be understood as the borrower's intended use of the loan.
[0046] S1012 extracts collection information of the target debtor from the target loan recipient's information based on the person template.
[0047] The target loan recipient corresponds to the target debt collector. The target loan recipient and the target debt collector can be the same person or different people. For example, the target debt collector can be a relative or friend of the target loan recipient. This application does not impose any restrictions on this.
[0048] Information about the target loan applicant includes, but is not limited to, the applicant's occupation, location, gender, age, education, relationship to the borrower (the applicant, immediate family member, spouse), and loan purpose.
[0049] S1013 selects m collection case information that match the collection information from the preset collection case library based on the collection information.
[0050] The preset collection case library includes information on n collection cases, where n and m are positive integers and n is greater than or equal to m.
[0051] In one embodiment, S1013 specifically includes the following steps: S10131-S10134.
[0052] S10131. Calculate the similarity between the collection information and each collection case information in the preset collection case library to obtain n similarity scores.
[0053] Among them, n similarity scores correspond one-to-one with n collection case information.
[0054] It should be noted that, in the embodiments of this application, the text content of both the collection information and the collection case information can be vectorized, and then the similarity between the vectors can be calculated to obtain n similarity scores.
[0055] S10132. Select the top m similarities from the n similarities in descending order of similarity.
[0056] In this embodiment, the n similarities are sorted from largest to smallest, and the top m similarities are selected from the n similarities based on practical experience.
[0057] S10133. Obtain the collection case information corresponding to the first m similarities, and get the information of m collection cases.
[0058] In this embodiment of the application, each similarity corresponds to a collection case information.
[0059] S1014 inputs the collection information and m collection case information into the preset large language model to obtain the virtual character of the target to be collected.
[0060] This application embodiment inputs collection information and m collection case information into a preset large language model to obtain a virtual character for the target collection object. The collection information is the collection information of the target collection object, making the constructed virtual character more closely resemble the target collection object.
[0061] In one embodiment, the process of constructing the preset collection case library includes: ac.
[0062] a. Obtain audio recordings of conversations from multiple historical debt collection cases.
[0063] b. Extract dialogue content from the audio recordings of each historical debt collection case to obtain multiple dialogue texts.
[0064] Multiple dialogue texts correspond one-to-one with multiple historical debt collection cases; c. Input multiple dialogue texts into a preset large language model to obtain multiple collection case information.
[0065] Multiple collection case information corresponds one-to-one with multiple dialogue texts.
[0066] It should be noted that step ac will be explained in detail below.
[0067] The collection case information includes relevant information about the collection personnel, the borrower, and the dialogue scenario. The collection personnel's information includes, but is not limited to, the compliance of their collection methods, whether they guide repayment, their emotional management skills, communication skills, empathy, logical communication, and professionalism. The borrower's information includes, but is not limited to, their tone, attitude, changes in attitude, occupation, location, gender, age group, education level, and relationship to the borrower (the borrower, immediate family member, spouse). The dialogue scenario includes, but is not limited to, the borrower's delinquency due to a short-term funding gap but with the intention to repay; the borrower forgetting to repay, and the borrower immediately addressing the issue after being reminded; the borrower being perfunctory, delaying, or making excuses to avoid repayment; the borrower being adamant and refusing to repay; a sudden family emergency causing inability to repay; and requests for an extension of the repayment period or a reduction in the amount of each payment.
[0068] This application embodiment stores multiple collection case information in a local database for use.
[0069] See Figure 2 , Figure 2 This is a schematic block diagram of a debt collector training device provided in an embodiment of this application. Corresponding to the above-described debt collector training method, this application also provides a debt collector training device. This debt collector training device includes a unit for executing the above-described debt collector training method, and can be configured in a terminal such as a desktop computer, tablet computer, or laptop computer. Specifically, the debt collector training device includes... Acquisition unit 201 is used to acquire the virtual character of the target debt collector; Dialogue unit 202 is used to conduct a simulated debt collection dialogue with the virtual character and obtain dialogue information; The input unit 203 is used to input the dialogue information and the first preset prompt word into the preset large language model to obtain the reference script corresponding to the debt collection simulation dialogue. The reference script is used to guide the debt collector to conduct a debt collection dialogue with the target debt collector. The first preset prompt word is used to instruct the preset large language model to analyze the reference script from the dialogue information. The preset large language model is used to analyze the reference script from the dialogue information based on the first preset prompt word.
[0070] In one embodiment, the input unit 203 is further configured to input the dialogue information and the second preset prompt word into the preset large language model to obtain a score corresponding to the debt collection simulation dialogue. The score is used to characterize the quality level of the debt collection simulation dialogue between the debt collector and the virtual character. The second preset prompt word is used to instruct the preset large language model to score the debt collection simulation dialogue. The preset large language model is also used to score the debt collection simulation dialogue based on the second preset prompt word.
[0071] In one embodiment, the input unit 203 is further configured to input the dialogue information and the third preset prompt word into the preset large language model to obtain summary information corresponding to the debt collection simulation dialogue. The summary information is used to characterize the shortcomings of the debt collector in the debt collection simulation dialogue. The preset large language model is further configured to summarize the dialogue information according to the third preset prompt word.
[0072] In one embodiment, the acquisition unit 201 is specifically used to define the person template of the target debt collector; Based on the person template, the collection information of the target loan object is extracted from the information of the target loan object, and the target loan object corresponds to the target collection object; Based on the collection information, m collection cases that match the collection information are selected from a preset collection case library. The preset collection case library includes n collection cases, where n and m are positive integers and n is greater than or equal to m. The collection information and the information of the m collection cases are input into the preset large language model to obtain the virtual character of the target collection object.
[0073] In one embodiment, the person template includes personality, occupation, gender, age, relationship with the borrower, and loan scenario.
[0074] In one embodiment, the acquisition unit 201 is further specifically used to calculate the similarity between the collection information and each collection case information in the preset collection case database, to obtain n similarity scores, wherein the n similarity scores correspond one-to-one with the n collection case information. Select the top m similarities from the n similarities in descending order of similarity. Obtain the collection case information corresponding to the first m similarities to obtain the m collection case information.
[0075] In one embodiment, the process of constructing the preset collection case library includes: Obtain audio recordings of conversations from multiple historical debt collection cases; The dialogue content is extracted from the voice recordings of each of the multiple historical collection cases to obtain multiple dialogue texts, and the multiple dialogue texts correspond one-to-one with the multiple historical collection cases. The multiple dialogue texts are input into the preset large language model to obtain multiple collection case information, and the multiple collection case information corresponds one-to-one with the multiple dialogue texts.
[0076] like Figure 3 As shown, this application provides a computer device including a processor 31, a communication interface 32, a memory 33, and a communication bus 34. The processor 31, the communication interface 32, and the memory 33 communicate with each other through the communication bus 34. The memory 33 is used to store computer programs. In one embodiment of this application, the processor 31, when executing the program stored in the memory 33, implements the control method for debt collector training provided in any of the foregoing method embodiments.
[0077] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program may be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.
[0078] Therefore, embodiments of this application also provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the collection personnel training method provided in any of the foregoing method embodiments.
[0079] The storage medium is a physical, non-transient storage medium, such as a USB flash drive, external hard drive, read-only memory (ROM), magnetic disk, or optical disk, or any other physical storage medium capable of storing program code. The computer-readable storage medium can be non-volatile or volatile.
[0080] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.
[0081] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.
[0082] The steps in the methods of this application embodiment can be adjusted, merged, or deleted according to actual needs. The units in the apparatus of this application embodiment can be merged, divided, or deleted according to actual needs. Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0083] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application.
[0084] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.
[0085] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Since these modifications and variations fall within the scope of the claims and their equivalents, this application also intends to include these modifications and variations.
[0086] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. A method for training debt collectors, characterized in that, include: Obtain the virtual avatar of the target debt collector; Engage in a simulated debt collection dialogue with the virtual character to obtain dialogue information; The dialogue information and the first preset prompt word are input into the preset large language model to obtain the reference script corresponding to the debt collection simulation dialogue. The reference script is used to guide the debt collector to conduct a debt collection dialogue with the target debt collector. The first preset prompt word is used to instruct the preset large language model to analyze the reference script from the dialogue information. The preset large language model is used to analyze the reference script from the dialogue information based on the first preset prompt word.
2. The method according to claim 1, characterized in that, After engaging in a simulated debt collection dialogue with the virtual character and obtaining dialogue information, the method further includes: The dialogue information and the second preset prompt word are input into the preset large language model to obtain the score corresponding to the debt collection simulation dialogue. The score is used to characterize the quality level of the debt collection simulation dialogue between the debt collector and the virtual character. The second preset prompt word is used to instruct the preset large language model to score the debt collection simulation dialogue. The preset large language model is also used to score the debt collection simulation dialogue based on the second preset prompt word.
3. The method according to claim 1 or 2, characterized in that, After engaging in a simulated debt collection dialogue with the virtual character and obtaining dialogue information, the method further includes: The dialogue information and the third preset prompt word are input into the preset large language model to obtain the summary information corresponding to the debt collection simulation dialogue. The summary information is used to characterize the shortcomings of the debt collector in the debt collection simulation dialogue. The preset large language model is also used to summarize the dialogue information based on the third preset prompt word.
4. The method according to claim 1 or 2, characterized in that, The process of obtaining the virtual character of the target debt collector includes: Define the person template for the target debt collector; Based on the person template, the collection information of the target loan object is extracted from the information of the target loan object, and the target loan object corresponds to the target collection object; Based on the collection information, m collection cases that match the collection information are selected from a preset collection case library. The preset collection case library includes n collection cases, where n and m are positive integers and n is greater than or equal to m. The collection information and the information of the m collection cases are input into the preset large language model to obtain the virtual character of the target collection object.
5. The method according to claim 4, characterized in that, The character template includes the character's personality, occupation, gender, age, relationship with the borrower, and loan scenario.
6. The method according to claim 4, characterized in that, The step of selecting m collection case information that match the collection information from a preset collection case database based on the collection information includes: Calculate the similarity between the collection information and each collection case information in the preset collection case database to obtain n similarity scores, and the n similarity scores correspond one-to-one with the n collection case information; Select the top m similarities from the n similarities in descending order of similarity. Obtain the collection case information corresponding to the first m similarities to obtain the m collection case information.
7. The method according to claim 4, characterized in that, The process of constructing the pre-set collection case library includes: Obtain audio recordings of conversations from multiple historical debt collection cases; The dialogue content is extracted from the voice recordings of each of the multiple historical collection cases to obtain multiple dialogue texts, and the multiple dialogue texts correspond one-to-one with the multiple historical collection cases. The multiple dialogue texts are input into the preset large language model to obtain multiple collection case information, and the multiple collection case information corresponds one-to-one with the multiple dialogue texts.
8. A debt collector training device, characterized in that, include: The acquisition unit is used to acquire the virtual character of the target debt collector. The dialogue unit is used to conduct simulated debt collection dialogues with the virtual character and obtain dialogue information. An input unit is used to input the dialogue information and a first preset prompt word into a preset large language model to obtain a reference script corresponding to the debt collection simulation dialogue. The reference script is used to guide the debt collector to conduct a debt collection dialogue with the target debtor. The first preset prompt word is used to instruct the preset large language model to analyze the reference script from the dialogue information. The preset large language model is used to analyze the reference script from the dialogue information based on the first preset prompt word.
9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, can implement the method as described in any one of claims 1 to 7.