Loan risk exhaustion method, apparatus and device, and medium

Abstract

The invention relates to the technical field of computers, and discloses a loan risk exhaustion method, device and equipment and a medium, and the method comprises the steps: determining a risk prompt item of a target customer based on historical data when a loan application of the target customer is received, and matching a target verbal skill corresponding to the risk prompt item in a preset verbal skill library; initiating a video exhaustion call to the target client based on the target verbal skill, and obtaining reply information and behavior information of the target client in a video exhaustion call process; according to the method, the corresponding verbal skill is matched through the historical data of the target customer to carry out video exhaustion, and the reply and the behavior in the exhaustion process are analyzed, so that the risk exhaustion report of the target customer is generated. Therefore, the exhaustion report of the customer is generated more efficiently and accurately, and the working efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a loan risk due diligence method, device, equipment and medium. Background Art

[0002] With the development of society, people's demand for funds is increasing, which has led to an increase in the loan business volume of the banking industry. Before banks issue loans, they need to conduct risk investigations on loan applicants in order to understand whether the customers have the corresponding repayment ability and willingness to repay.

[0003] Normally, when conducting risk due diligence on loan customers, due diligence is often conducted offline with real people to clarify the possible risks of the loan and decide whether to issue the loan. However, when conducting risk due diligence in this way, since it mainly relies on manual experience to conduct investigations, the risk identification ability is relatively poor. Summary of the invention

[0004] In view of this, the present invention provides a loan risk due diligence method, apparatus, computer equipment and storage medium to solve the problem of relatively poor risk identification ability during risk due diligence.

[0005] In a first aspect, the present invention provides a loan risk due diligence method, the method comprising:

[0006] When receiving a loan application from a target customer, determining the risk warning items of the target customer based on historical data, and matching the target speech corresponding to the risk warning items in a preset speech library;

[0007] Initiate video due diligence on target customers based on the target words, and obtain response information and behavior information of target customers during the video due diligence process;

[0008] The response information and behavior information are analyzed based on the verification data and behavior standards to generate a risk due diligence report on the target customer.

[0009] This method determines the sales pitch for the target customer through the target customer's historical data, and then conducts video due diligence on the target customer based on the sales pitch. It also analyzes the response information and behavior information obtained during the due diligence process in combination with verification data and behavior standards. This can provide a more accurate understanding of the potential risks of the customer, and can also assist staff in conducting risk due diligence, thereby simplifying workload and improving work efficiency.

[0010] In an optional implementation, the determining the risk warning items of the target customer based on historical data includes:

[0011] Obtaining a preset screening table, wherein the preset screening table includes: a plurality of items to be screened;

[0012] Based on the historical data of the target customers, determine the historical information corresponding to each item to be screened in the preset screening table;

[0013] The risk warning items of the target customer are determined according to the historical information corresponding to the items to be screened.

[0014] In this implementation, the historical information of each screening item in the historical data of the target customer is determined by pre-setting the items to be screened in the screening table, thereby determining the risk warning items of the target customer. This can ensure the comprehensiveness of the screening of the user's potential risk warning items and avoid omissions, while making it convenient for the staff to update the items that need to be screened based on actual conditions.

[0015] In an optional implementation manner, the target customer's response information and behavior information during the video due diligence process is obtained in the following manner:

[0016] Acquire voice data and video data during the video due diligence process, parse the voice data, and obtain response information during the video due diligence process;

[0017] Based on the video data and voice data, behavioral information corresponding to multiple behavioral indicators of the target customer during the video due diligence process is determined, and the multiple behavioral indicators include: answer pause time, answer concentration, answer matching, and whether there are prompts in the surrounding environment.

[0018] In this embodiment, the voice data in the video due diligence process is analyzed to obtain the response information in the due diligence process. At the same time, the specific information under multiple behavioral indicators in the user due diligence process is determined based on the voice data and video data. The user's behavioral information and response information in the due diligence process can be understood more accurately to ensure the accuracy of subsequent risk analysis.

[0019] In an optional implementation, determining the behavior information corresponding to multiple behavior indicators of the target customer during the video due diligence process based on the video data and voice data includes:

[0020] Counting the pause time of the target customer's response during the due diligence process based on the voice data, and comparing the response information corresponding to the voice data with the due diligence questions to determine the response matching degree;

[0021] Determining the focus of the response according to whether the target customer has abnormal behavior when responding in the video data;

[0022] Determine whether there are any prompts in the surrounding environment of the target customer during the video due diligence process based on the video data and the voice data.

[0023] This implementation method determines the customer's response matching degree, response pause time, response concentration, and whether there are prompts in the surrounding environment and other behavioral information through the specific circumstances of the customer's response during the due diligence process. It can more comprehensively and accurately determine whether there are abnormalities in the user's response, thereby improving the accuracy of subsequent risk analysis.

[0024] In an optional implementation, the verification data includes: public data and record data in a database;

[0025] The analysis of the response information and behavior information based on the verification data and behavior standards to generate a risk due diligence report on the target customer includes:

[0026] Verify the target customer's reply information based on public data and record data in the database to determine the authenticity of the reply information;

[0027] Compare the behavior information corresponding to multiple behavior indicators with the corresponding behavior standards to determine the abnormality of the response behavior during the due diligence process;

[0028] Based on the authenticity of the response information and the abnormality of the response behavior, a risk due diligence report on the target customer is generated.

[0029] In this implementation mode, the customer's response information is verified through specific verification data to determine the authenticity of the response content. At the same time, the behavioral information under multiple behavioral indicators is compared with the specific behavioral standards to determine the degree of abnormality. By jointly analyzing the two, the potential risks of the target customers can be determined more comprehensively and accurately, and a comprehensive and accurate risk due diligence report can be generated.

[0030] In an optional implementation, after obtaining the risk warning item of the target customer, the method further includes:

[0031] Determine the risk level corresponding to the risk warning item;

[0032] When the risk level is the first preset level, a document review request is sent to the user according to the target customer's loan application, so that the user provides the document to be reviewed, and a risk due diligence report of the target customer is generated based on the document to be reviewed provided by the user;

[0033] When the risk level is a second preset level, the step of initiating video due diligence on the target customer based on the target speech is executed, and the risk level represented by the second preset level is higher than the first preset level.

[0034] This implementation method implements the video due diligence step when the risk level of the risk warning item is too high. When the risk level is low, only the materials to be reviewed provided by the target customer are verified. This can avoid excessive investigation, resulting in a lengthy loan process and affecting customer experience.

[0035] In an optional implementation, the initiating a video due diligence on the target customer based on the target speech includes:

[0036] A virtual digital person is generated, and a video due diligence is initiated to the target customer based on the target speech through the virtual digital person.

[0037] In this embodiment, a virtual digital human is used to conduct video due diligence on the user, which can improve the user experience during the due diligence process and make the user's response information and behavior information closer to the actual due diligence.

[0038] In a second aspect, the present invention provides a loan risk due diligence device, the device comprising:

[0039] A speech matching module is used to determine the risk warning items of the target customer based on historical data when receiving a loan application from the target customer, and match the target speech corresponding to the risk warning item in a preset speech library;

[0040] A video due diligence module is used to initiate a video due diligence on a target customer based on the target speech, and obtain the target customer's response information and behavior information during the video due diligence process;

[0041] The information analysis module is used to analyze the response information and behavior information based on the verification data and behavior standards to generate a risk due diligence report on the target customer.

[0042] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the loan risk due diligence method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0043] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the loan risk due diligence method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 is a flowchart of a loan risk due diligence method according to an embodiment of the present invention;

[0046] Figure 2 is a flow chart of another loan risk due diligence method according to an embodiment of the present invention;

[0047] Figure 3 This is an example diagram of an intelligent-driven virtual digital human operation process according to an embodiment of the present invention;

[0048] Figure 4 This is an example diagram of the operation process of a real-person driven virtual digital human according to an embodiment of the present invention;

[0049] Figure 5 is an example diagram of a key technology of a virtual digital human according to an embodiment of the present invention;

[0050] Figure 6 This is an example diagram of a loan approval process based on a virtual digital person according to an embodiment of the present invention;

[0051] Figure 7 This is a structural block diagram of a loan risk due diligence device according to an embodiment of the present invention;

[0052] Figure 8 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0054] With the development of society, people's demand for funds is increasing, which has led to an increase in the loan business volume of the banking industry. Before banks issue loans, they need to conduct risk investigations on loan applicants in order to understand whether the customers have the corresponding repayment ability and willingness to repay.

[0055] Normally, when conducting risk due diligence on loan customers, due diligence is often conducted offline with real people to clarify the possible risks of the loan and decide whether to issue the loan. However, when conducting risk due diligence in this way, since it mainly relies on manual experience to conduct investigations, the risk identification ability is relatively poor.

[0056] To this end, an embodiment of the present invention provides a loan risk due diligence method, which determines the sales pitch for the target customer through the target customer's historical data, and then conducts video due diligence on the target customer based on the sales pitch. The response information and behavior information obtained during the due diligence process are analyzed in combination with verification data and behavior standards. This can more accurately understand the potential risks of the customer, and at the same time assist staff in conducting risk due diligence, simplifying the workload and improving work efficiency.

[0057] According to an embodiment of the present invention, an embodiment of a loan risk due diligence method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0058] In this embodiment, a loan risk due diligence method is provided, which can be used for the risk due diligence of the above-mentioned bank loan. Figure 1 is a flow chart of a loan risk due diligence method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0059] Step S101, when a loan application from a target customer is received, the risk warning items of the target customer are determined based on historical data, and the target speech corresponding to the risk warning items is matched in a preset speech library.

[0060] For those who want to apply for loan business, they can apply for loan business independently through relevant terminals of the bank, such as self-service business processing machines, mobile terminals, etc. When receiving a loan request initiated by a customer, the historical data of the customer can be called up in the internal historical database of the bank, which can include, for example, historical deposit business, loan repayment status, salary income status and other related information. In this way, it can be determined whether the user has relevant risk items in history, such as risk items such as loan expectations, income and loan demand not meeting the requirements.

[0061] For these risk items, the system will preset a corresponding script library. At this time, the corresponding target script can be matched according to the specific risk items, so that the corresponding target script can be used to conduct risk investigation on the user in the future. In some cases, there are no risk items in the customer's historical data, that is, the customer is a good user. At this time, the corresponding basic script can be matched to conduct due diligence on some basic projects, or only the customer can be asked for the information that needs to be verified, and the corresponding information of the customer can be verified without subsequent due diligence.

[0062] Step S102, initiating a video due diligence on the target customer based on the target speech, and obtaining the target customer's response information and behavior information during the video due diligence process.

[0063] After obtaining the target script corresponding to the target customer, a video due diligence can be initiated to the customer through a virtual person, that is, due diligence communication with the customer on relevant issues in the target script can be carried out through video. During the communication, the content of the user's reply can be analyzed, and communication on various issues can be carried out according to the specific content of the target script. The specific execution of this part can be achieved through preset semantic recognition, knowledge base call and other related technologies. During the due diligence process, the specific reply information and behavior information of the customer during the due diligence communication process can be obtained, where the reply information is the specific content of the customer's reply, and the behavior information refers to the customer's behavior when replying, etc., which is used to subsequently determine whether the customer has abnormal behavior, so as to assess the corresponding risks.

[0064] In an example, suppose the user's risk warning item is: customer credit overdue, and the corresponding historical information is that the overdue is not malicious, the overdue amount is <10,000, and it has been settled.

[0065] At this point, the target speech that the system matches in the target speech library can be as follows:

[0066] 1. The current system has detected that one of your loans is overdue. Are you aware of this?

[0067] 2. Do you remember which bank the loan was from?

[0068] 3. What was the overdue amount at that time?

[0069] 4. How many days was the payment overdue?

[0070] 5. What is the reason for the overdue payment?

[0071] 6. Did the bank remind you to repay the loan at that time? (The branch of forgetting to repay refers to the follow-up issues when the customer forgets to repay and causes overdue payment).

[0072] 7. How did the bank remind you to repay the loan? (The branch where the bank reminds you to repay, i.e. the follow-up questions when the customer's bank reminds you to repay the loan).

[0073] 8. Have you paid it back now?

[0074] 9. Please provide a screenshot of your repayment. (Paid, that is, the customer has repaid the overdue loan).

[0075] In one example, a series of inquiries can be made to the customer through a virtual image regarding the above-mentioned words to complete the entire video due diligence process, while the response information and behavior information during the video due diligence process can be collected.

[0076] It should be noted that in the relevant embodiments of the present invention, the user will be informed in advance when collecting the user's behavior information and response information, and the relevant information will be collected only when the user agrees to authorize. Therefore, there is no issue of infringing on the privacy of citizens. At the same time, the collected information is only used for subsequent risk analysis, and the relevant data will be deleted after the risk assessment report is generated.

[0077] Step S103, analyzing the response information and behavior information based on the verification data and behavior standards, and generating a risk due diligence report for the target customer.

[0078] After obtaining the response information and behavior information during the due diligence process, the response information needs to be verified based on the corresponding verification data to determine whether the content of the customer's response is true. For the behavior information, it is necessary to compare the behavior information according to the corresponding behavior standards to determine whether the customer has potential risks such as lying and coercion in the response process. After analyzing these two perspectives, a specific risk due diligence report can be obtained, which can include the accuracy of the customer's answers to different questions, as well as the customer's abnormal behavior during the response process, and give the corresponding risk assessment conclusion, that is, whether the customer has potential risks and how high the risk is.

[0079] The loan risk due diligence method provided in this embodiment determines the sales pitch for the target customer through the target customer's historical data, and then conducts video due diligence on the target customer based on the sales pitch. The response information and behavior information obtained during the due diligence process are analyzed in combination with verification data and behavior standards. This can more accurately understand the potential risks of the customer, and can also assist staff in conducting risk due diligence, simplifying workload and improving work efficiency.

[0080] According to an embodiment of the present invention, another loan risk due diligence method embodiment is provided, which can be used for the risk due diligence of the above-mentioned bank loan. Figure 2 is a flow chart of another loan risk due diligence method according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0081] Step S201, when a loan application from a target customer is received, the risk warning items of the target customer are determined based on historical data, and the target speech corresponding to the risk warning items is matched in a preset speech library.

[0082] Specifically, in the above step S201, the risk warning items of the target customer are determined based on the historical data, including:

[0083] Step S201 - 1 , obtaining a preset screening table, the preset screening table includes: a plurality of items to be screened.

[0084] The preset screening table can be understood as the commonly used screening items when the bank analyzes the historical data of loan customers, which is used to evaluate the risks represented by the customer's historical data. For example, the items to be screened in the preset screening item table may include: overdue loan items, income details items, assets under the name and other items that need to be screened.

[0085] Step S201 - 2 : based on the historical data of the target customer, determine the historical information corresponding to each to-be-screened item in the preset screening table.

[0086] Combine the various items to be screened in the preset screening table and the specific historical data to determine the historical information of each screening item. For example, determine based on historical data whether the target customer has a loan arrears; whether the target customer's income can meet the loan repayment; whether the assets under the target customer's name are sufficient for a mortgage loan, etc. The specific historical information corresponding to each item to be screened.

[0087] Step S201 - 3 , determining the risk warning items of the target customer according to the historical information corresponding to each item to be screened.

[0088] After determining the historical information corresponding to each of the user's screening items, you can determine whether the corresponding screening items are risk items based on the specific historical information. For example, if the user has historical loan expectations, then the loan expectations are a risk item; if the user's historical income does not meet the loan repayment requirements, then the income situation is a risk item, and so on, to determine the user's specific risk items, where there can be multiple risk items, and different types of risks have different weights, which are used to measure the overall risk of the customer in history, so as to match more rigorous and detailed scripts in the script library.

[0089] In one example, after obtaining the risk warning items of the target customer, the method further includes:

[0090] Step S201 - a, determining the risk level corresponding to the risk warning item.

[0091] It can be understood that after obtaining the user's specific risk warning items, the specific risk level corresponding to the historical risk items can be determined based on the specific number of risk warning items, risk weights, etc., so as to decide whether to conduct video due diligence on the target customers. For example, when the risk level is low, only the corresponding verification information needs to be provided. When the risk level is high, video due diligence is required. When the risk level is extremely high, a real person is required to conduct an offline review.

[0092] Step S201-b, when the risk level is the first preset level, a document review request is sent to the user according to the target customer's loan application, so that the user provides the document to be reviewed, and a risk due diligence report of the target customer is generated based on the document to be reviewed provided by the user.

[0093] When the risk level represented by the risk warning item is low, it is the first preset level. At this time, it means that the risk represented by the user's historical data is low, and there is no need to conduct video due diligence. You only need to ask the user for the information that needs to be verified, and generate a report based on the specific verification results of the information. The report mainly shows which information contains incorrect information.

[0094] Step S201-c, when the risk level is the second preset level, execute the step of initiating video due diligence on the target customer based on the target speech, and the risk level represented by the second preset level is higher than the first preset level.

[0095] When the risk level is higher, that is, the second preset level, it is necessary to conduct a video due diligence on the user to determine whether the user has potential risks based on the user's specific responses and behavioral information during the video due diligence process.

[0096] Specifically, in one example, after obtaining the risk warning items and risk level of the user, the specific face-to-face review process for the user can be determined. There are four main types of face-to-face review processes:

[0097] 1. Data review, i.e. requesting the user to review the data that needs to be verified, refer to the above step S201-b.

[0098] 2. Telephone verification, that is, initiating a telephone due diligence on the user, that is, conducting a line analysis based on the user's responses during the telephone communication, without the need to analyze the specific details of the user's behavior.

[0099] 3. Video verification, that is, initiating video due diligence on users and determining the potential risks of users by analyzing their responses and behaviors, which is the main application scenario of the present invention.

[0100] 4. Offline due diligence, that is, when the risk level is extremely high, it is necessary to notify relevant bank employees to conduct real-person verification to avoid loan risks.

[0101] The risk levels corresponding to the above four face-to-face review processes increase step by step. That is, when the risk level represented by the risk warning item is the lowest, only document review is required. When the risk level is the highest, offline due diligence is required.

[0102] In one example, when conducting offline risk due diligence in person, if multiple risk warning items are designed, the risk warnings need to be determined according to the principle of minimum consolidation to determine the comprehensive risk, and the process of real-person review can be determined according to the level of risk. Specifically, for low risk, the operation manager will review alone; for medium risk, the operation manager-risk manager will review; for high risk, the operation manager-risk manager-independent approval officer will review.

[0103] Step S202, initiating video due diligence on target customers based on target sales pitch.

[0104] Specifically, we initiate video due diligence on target customers based on the target words, including:

[0105] Generate a virtual digital person and use the virtual digital person to initiate video due diligence on target customers based on target scripts.

[0106] It can be understood that when initiating video due diligence on users, a virtual digital person can be generated, and a series of questions can be asked to the customer based on the image of the virtual digital person according to the relevant words, and corresponding responses can be made according to the user's replies, so that the status of the customer during the due diligence is closer to the real person review, ensuring that the customer's performance during the due diligence process is close to the actual real person video due diligence. Figure 3 The figure shows an example diagram of the operation process of an intelligent-driven virtual digital human according to an embodiment of the present invention.

[0107] The virtual digital human exchanges information with the user through the terminal, recognizes and perceives the information input by the user, that is, the user's voice data and video data, and thus performs intelligent analysis and decision-making. That is, based on the target words determined above, the corresponding reply text is selected, and the text is used to generate task voice through a preset voice program. At the same time, combined with the animation generation program, the corresponding character animation is adaptively generated, and the character animation and voice are synthesized and displayed in audio and video. The terminal interacts with the user's itinerary, thereby executing the information interaction during the video due diligence process.

[0108] In another case, when a real person is needed to conduct due diligence on a user, a virtual digital person can also be used to conduct due diligence remotely, such as Figure 4 The figure shows an example diagram of the operation process of a real-person driven virtual digital human according to an embodiment of the present invention.

[0109] When a virtual digital person generated based on a real person interacts with a user, the real person can directly output the corresponding questions through audio and video monitoring data. When outputting information, the voice data is obtained by recording the real person's voice, and an animated image is generated through animation capture and migration. The two are combined to perform audio and video synthesis and display to obtain output information, so as to interact with the user and perform remote due diligence.

[0110] For virtual digital humans, there are 2D digital humans and 3D digital humans. The difference between the two lies mainly in task generation, such as Figure 5 , which is an example diagram of key technologies of a virtual digital human according to an embodiment of the present invention.

[0111] When generating digital humans, 2D digital humans do not need character modeling and can directly call the set image. 3D digital humans need to obtain specific character images through character modeling. When expressing, character expression is required, including voice generation and animation generation. Animation generation includes two major parts: driving (action generation) and rendering. "Two verticals" refer to 2D and 3D digital humans. 3D digital humans need to use additional three-dimensional modeling technology to generate digital images. The information dimension increases and the amount of calculation required is greater. When synthesizing the display, the display is performed through terminal display technology. When performing recognition, perception, analysis and decision-making, voice and semantic recognition, face recognition, action recognition, knowledge base, dialogue management and other technologies are mainly used.

[0112] Step S203, obtaining the target customer's response information and behavior information during the video due diligence process.

[0113] Specifically, in step S203, it includes:

[0114] Step S203-1, obtaining voice data and video data in the video due diligence process, parsing the voice data, and obtaining response information in the video due diligence process.

[0115] It can be understood that in the process of conducting video due diligence on customers based on specific words, the user's voice data can be analyzed to obtain their specific response information to the due diligence questions, so as to ask other subsequent questions based on the response information. Figure 1 Examples of specific response language at various locations in the illustrated embodiment, with options for subsequent due diligence questions for different response information.

[0116] Step S203-2, based on the video data and voice data, determine the behavior information corresponding to multiple behavior indicators of the target customer during the video due diligence process, and the multiple behavior indicators include: response pause time, response concentration, response matching, and whether there are prompts in the surrounding environment.

[0117] In addition to analyzing voice data, voice data and video data can also be combined to jointly determine the customer's behavioral information during the video due diligence process. For example, the specific responses in the voice can be analyzed and compared with the real data. At the same time, the fluency of the user's response and whether there are abnormal actions can be combined to obtain more accurate and specific behavioral information. For example, the pause time of the response in the voice data can be counted, the user's concentration when responding can be judged through video data, whether there are prompts in the surrounding environment, and the matching degree between the response information and the actual questions can be analyzed. Specific behavioral information can then be used to judge the reliability of the user's response during the due diligence process based on the set evaluation rules.

[0118] Furthermore, based on the video data and voice data, the behavior information corresponding to multiple behavior indicators of the target customer during the video due diligence process is determined, including:

[0119] Step S203-2-1, count the pause time of the target customer's response during the due diligence process based on the voice data, and compare the response information corresponding to the voice data with the due diligence questions to determine the response matching degree.

[0120] It can be understood that the pause time of the user when answering the question is counted to obtain the answer pause time, which can, to a certain extent, represent whether the user truly understands the specific situation corresponding to the question.

[0121] By comparing the response information with the due diligence questions, we can determine whether the user's answer is targeted, so that when the user avoids answering the question by changing the subject, we can understand the user's true intention in time, so as to perceive the corresponding potential risks in subsequent analysis.

[0122] The matching degree of the answer and the pause time can reflect the user's true psychological state during the due diligence process from a certain perspective. For example, when asking about the overdue amount, if the user pauses for too long or too short, it may indicate that the user's answer is abnormal. If the pause time is too long, it may indicate that the user does not understand the specific overdue situation at the time. If the answer time is too short, it means that the user may have prepared for the question in advance, resulting in doubts about the authenticity of the answer.

[0123] Step S203-2-2, determining the focus of the response based on whether the target customer has abnormal behavior when responding in the video data.

[0124] In one example, abnormal behavior may be looking around, checking the phone multiple times, asking others, etc. These behaviors may indicate that the user does not understand his or her actual credit status, and are measured here by the concentration of the response.

[0125] Step S203-2-3, determine whether there are any prompts in the surrounding environment of the target customer during the video due diligence process based on the video data and voice data.

[0126] For example, video data can be used to detect whether there are other people or reminders around the customer, and voice data can be combined to determine whether there are other people around the customer when he or she is answering, etc. This will facilitate subsequent overall analysis of the authenticity of the customer's answer.

[0127] In one example, the specific behavior information of the customer can be set in combination with psychology or actual due diligence experience. The main purpose of collecting behavior information is to analyze whether the user's response is based on his or her own understanding, that is, the authenticity of the response. The setting of specific behavior indicators can be set in combination with actual conditions.

[0128] Step S204, analyzing the response information and behavior information based on the verification data and behavior standards, and generating a risk due diligence report for the target customer.

[0129] Specifically, the verification data includes: public data and record data in the database.

[0130] Public data may include: the customer's public credit status, name, gender, companies under his name, assets and other publicly available information, and recorded data may be the customer's historical financial data within the bank. It should be noted that the customer data used in this embodiment are all recorded data within the bank or publicly available information, so there is no violation of citizens' privacy.

[0131] In step S204, the response information and behavior information are analyzed based on the verification data and behavior standards to generate a risk due diligence report for the target customer, including:

[0132] Step S204-1, based on the public data and the record data in the database, verify the reply information of the target customer to determine the authenticity of the reply information.

[0133] The user's reply information is verified based on specific internal bank record data or available public data to determine the authenticity of the user's reply information.

[0134] Step S204-2, comparing the behavior information corresponding to the multiple behavior indicators with the corresponding behavior standards to determine the abnormality of the response behavior during the due diligence process.

[0135] Compare each behavior indicator with the corresponding behavior standard to determine the abnormality of the user's response behavior. For example, the behavior standard for the pause time of the response is 1-5 seconds, and the average pause time for each question when the user is responding is about 10 seconds, which means that the behavior information is abnormal. Similarly, the behavior information under each behavior indicator has a corresponding specific behavior standard, which is compared with the specific behavior standard to determine the abnormality of each behavior indicator. The setting of specific behavior standards can refer to actual banking business experience and is not limited here.

[0136] Step S204-3, generating a risk due diligence report for the target customer based on the authenticity of the reply information and the abnormality of the reply behavior.

[0137] After obtaining the authenticity of the customer's response information and the abnormality of each response behavior during the entire due diligence process, we can give an overall risk assessment result based on the preset assessment rules, such as setting different risk weights for the abnormality of different behaviors or mismatched response information. At the same time, in the risk due diligence report, we can specifically point out which issues of the customer should be abnormal or the response information does not match, so that the staff can approve the loan issuance based on the due diligence report.

[0138] The loan risk due diligence method provided in the embodiment of the present invention determines the sales pitch for the target customer through the historical data of the target customer, and then conducts video due diligence on the target customer based on the sales pitch, and analyzes the response information and behavior information obtained during the due diligence process in combination with verification data and behavior standards, thereby more accurately understanding the potential risks of the customer and assisting staff in conducting risk due diligence, simplifying workload and improving work efficiency.

[0139] In order to facilitate understanding of the specific content of the above-mentioned invention embodiment, an example diagram of a loan approval process based on a virtual digital person according to an embodiment of the invention is shown as follows: Figure 6 shown.

[0140] First, the due diligence component triggers the virtual person to conduct due diligence based on the customer's loan application. Before conducting due diligence, the risk SOP process component is needed to determine the corresponding risk speech process table based on the customer's historical risk prompts, so as to conduct subsequent risk due diligence, that is, based on the risk speech library, ask questions to the customer according to the standardized process.

[0141] Specifically, the virtual digital human component can be used to execute subsequent due diligence processes, provide remote video interaction capabilities with customers, enable real-time audio and screen recording, and provide commonly used customer uploaded information and automatic OCR capabilities.

[0142] When conducting risk due diligence, the voice analysis component and customer behavior analysis component are used to analyze the customer's voice and conduct behavioral analysis on the customer's interactive behavior in the video.

[0143] The speech analysis component can analyze the customer's speech to form structured data, and extract keywords to obtain key information. The first step is to convert all the questions asked by the account manager or AI and the customer's answers into text. The second step is to match and extract the corresponding sensitive words. The third step is to match and store the conclusions output by the customer behavior analysis component with the corresponding questions and answers. The fourth step is to match and store the conclusions output by the data cross-validation component involving customer answers with the corresponding questions and answers.

[0144] The customer behavior analysis component can analyze the customer's behavior and responses in the video, and issue risk warnings for specific behaviors. The specific analysis methods for behaviors and responses mainly include analysis of the pause time of customer responses, analysis of the accuracy of customer responses (number of repetitions and matching degree of correct answers), analysis of the customer's surrounding environment (whether there are external sound prompts), and analysis of the number of times the customer has modified the information they filled out (if there is additional information involved).

[0145] Then, the customer information and reply content are cross-verified through the data cross-verification component. For example, the main verification is: the overdue loan bank and overdue amount in the customer's reply are cross-verified with the data in the credit report. The bank repayment reminder method in the customer's reply is cross-verified with the repayment reminder method of the bank corresponding to the system configuration.

[0146] The data cross-verification of application materials mainly includes the customer's loan amount, monthly income, work unit, work address, mobile phone number, etc., and the corresponding third-party data for data verification. It should be noted that these data are called with the user's authorization, so there is no violation of personal privacy.

[0147] The underlying layer of data verification is the big data model. The model is used for calculation to determine the customer's true intention. The true intention is determined based on the touching of risk indicators. The calculation results of the verification model are used to feedback the customer's true intention. The specific big data model can refer to the relevant specific technologies, which will not be elaborated here.

[0148] Finally, after completing the above steps, the risk prediction component can be used to conduct a comprehensive analysis of the specific information collected and the specific verification results to obtain the user's risk due diligence report. When conducting a comprehensive analysis, the user behavior is mainly analyzed to determine whether the user's loan intention is real and whether the relevant response information is accurate. Specifically, different impact weights can be set for different behaviors or responses to obtain a global user risk assessment result.

[0149] In this embodiment, a loan risk due diligence device is also provided, which is used to implement the above embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0150] This embodiment provides a loan risk due diligence device, such as Figure 7 As shown, including:

[0151] The speech matching module 401 is used to determine the risk warning items of the target customer based on historical data when receiving a loan application from the target customer, and match the target speech corresponding to the risk warning item in the preset speech library;

[0152] The video due diligence module 402 is used to initiate a video due diligence with the target customer based on the target speech, and obtain the target customer's response information and behavior information during the video due diligence process;

[0153] The information analysis module 403 is used to analyze the response information and behavior information based on the verification data and behavior standards to generate a risk due diligence report for the target customer.

[0154] In some optional implementations, the speech matching module 401, when determining the risk warning items of the target customer based on the historical data, includes:

[0155] Obtain a preset screening form, which includes: multiple items to be screened;

[0156] Based on the historical data of the target customers, determine the historical information corresponding to each item to be screened in the preset screening table;

[0157] Determine the risk warning items for target customers based on the historical information corresponding to each item to be screened.

[0158] In some optional implementations, in the video due diligence module 402, the target customer's response information and behavior information during the video due diligence process are obtained in the following manner:

[0159] Acquire voice data and video data during the video due diligence process, parse the voice data, and obtain response information during the video due diligence process;

[0160] Based on video data and voice data, the behavioral information corresponding to multiple behavioral indicators of target customers during the video due diligence process is determined. The multiple behavioral indicators include: answer pause time, answer concentration, answer matching, and whether there are prompts in the surrounding environment.

[0161] In some optional implementations, the video due diligence module 402, when determining the behavior information corresponding to multiple behavior indicators of the target customer during the video due diligence process based on the video data and the voice data, includes:

[0162] Count the pause time of target customers’ responses during the due diligence process based on voice data, and compare the response information corresponding to the voice data with the due diligence questions to determine the matching degree of the responses;

[0163] Determine the focus of the response based on whether the target customer has abnormal behavior when responding in the video data;

[0164] Determine whether there are any prompts in the target customer's surrounding environment during the video due diligence process based on video data and voice data.

[0165] In some optional implementations, the verification data includes: public data and record data in a database. The information analysis module 403, when analyzing the response information and behavior information based on the verification data and behavior standards to generate a risk due diligence report for the target customer, includes:

[0166] Verify the target customer's response information based on public data and record data in the database to determine the authenticity of the response information;

[0167] Compare the behavior information corresponding to multiple behavior indicators with the corresponding behavior standards to determine the abnormality of the response behavior during the due diligence process;

[0168] Generate a risk due diligence report on the target customer based on the authenticity of the response information and the abnormality of the response behavior.

[0169] In some optional implementations, after obtaining the risk warning items of the target customer, the speech matching module 401 is further used to:

[0170] Determine the risk level corresponding to the risk warning item;

[0171] When the risk level is the first preset level, a document review request is sent to the user according to the target customer's loan application, so that the user provides the documents to be reviewed, and a risk due diligence report of the target customer is generated based on the documents to be reviewed provided by the user;

[0172] When the risk level is the second preset level, the video due diligence module 402 is called to execute the step of initiating video due diligence on the target customer based on the target speech, and the risk level represented by the second preset level is higher than the first preset level.

[0173] In some optional implementations, the video due diligence module 402, when initiating a video due diligence to a target customer based on the target speech, includes:

[0174] Generate a virtual digital person and use the virtual digital person to initiate video due diligence on target customers based on target scripts.

[0175] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0176] The loan risk due diligence device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0177] The embodiment of the present invention also provides a computer device having the above Figure 7 The xxxx device shown.

[0178] See also Figure 8 , Figure 8 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 8 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 8 A processor 10 is taken as an example.

[0179] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0180] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0181] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0182] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0183] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 8 The example of connecting through bus is taken in the following.

[0184] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0185] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0186] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A loan risk due diligence method, characterized in that: The method comprises: When receiving a loan application from a target customer, determining the risk warning items of the target customer based on historical data, and matching the target speech corresponding to the risk warning items in a preset speech library; Initiate video due diligence on target customers based on the target words, and obtain response information and behavior information of target customers during the video due diligence process; The response information and behavior information are analyzed based on the verification data and behavior standards to generate a risk due diligence report on the target customer.

2. The method according to claim 1, characterized in that The risk warning items of the target customer determined based on historical data include: Obtaining a preset screening table, wherein the preset screening table includes: a plurality of items to be screened; Based on the historical data of the target customers, determine the historical information corresponding to each item to be screened in the preset screening table; The risk warning items of the target customer are determined according to the historical information corresponding to the items to be screened.

3. The method according to claim 1, characterized in that The method for obtaining the target customer's response information and behavior information during the video due diligence process is as follows: Acquire voice data and video data during the video due diligence process, parse the voice data, and obtain response information during the video due diligence process; Based on the video data and voice data, behavioral information corresponding to multiple behavioral indicators of the target customer during the video due diligence process is determined, and the multiple behavioral indicators include: answer pause time, answer concentration, answer matching, and whether there are prompts in the surrounding environment.

4. The method according to claim 3, characterized in that Determining the behavior information corresponding to multiple behavior indicators of the target customer during the video due diligence process based on the video data and voice data includes: Counting the pause time of the target customer's response during the due diligence process based on the voice data, and comparing the response information corresponding to the voice data with the due diligence questions to determine the response matching degree; Determining the focus of the response according to whether the target customer has abnormal behavior when responding in the video data; Determine whether there are any prompts in the surrounding environment of the target customer during the video due diligence process based on the video data and the voice data.

5. The method according to claim 3, characterized in that: The verification data includes: public data and record data in the database; The analysis of the response information and behavior information based on the verification data and behavior standards to generate a risk due diligence report on the target customer includes: Verify the target customer's reply information based on public data and record data in the database to determine the authenticity of the reply information; Compare the behavior information corresponding to multiple behavior indicators with the corresponding behavior standards to determine the abnormality of the response behavior during the due diligence process; Based on the authenticity of the response information and the abnormality of the response behavior, a risk due diligence report on the target customer is generated.

6. The method according to claim 1, characterized in that After obtaining the risk warning item of the target customer, the method further includes: Determine the risk level corresponding to the risk warning item; When the risk level is the first preset level, a document review request is sent to the user according to the target customer's loan application, so that the user provides the document to be reviewed, and a risk due diligence report of the target customer is generated based on the document to be reviewed provided by the user; When the risk level is a second preset level, the step of initiating video due diligence on the target customer based on the target speech is executed, and the risk level represented by the second preset level is higher than the first preset level.

7. The method according to claim 1, characterized in that The initiating video due diligence on the target customer based on the target speech includes: A virtual digital person is generated, and a video due diligence is initiated to the target customer based on the target speech through the virtual digital person.

8. A loan risk due diligence device, characterized in that: The device comprises: A speech matching module is used to determine the risk warning items of the target customer based on historical data when receiving a loan application from the target customer, and match the target speech corresponding to the risk warning item in a preset speech library; A video due diligence module is used to initiate a video due diligence on a target customer based on the target speech, and obtain the target customer's response information and behavior information during the video due diligence process; The information analysis module is used to analyze the response information and behavior information based on the verification data and behavior standards to generate a risk due diligence report on the target customer.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the loan risk due diligence method according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the loan risk due diligence method according to any one of claims 1 to 7.

Images