Call-out report generation method and device, electronic equipment and storage medium

Abstract

The invention provides an exhaustion report generation method and device, electronic equipment and a storage medium, and relates to the technical field of data processing. The method comprises the following steps: acquiring object transaction data of a plurality of candidate objects; performing object identification on the object transaction data based on a preset identification strategy, and screening out at least one target object from the plurality of candidate objects; determining a target exhaustion verbal skill template matched with each target object in the at least one target object; for each target object, initiating a call outbound to the target object based on the target exhaustion call skill template to obtain an exhaustion voice dialogue of the target object; and generating a target exhaustion report of the target object based on the exhaustion voice dialogue. According to the method, full-link automatic customer full-duty investigation can be realized.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a due diligence report generation method, apparatus, electronic device, and storage medium. Background Technology

[0002] As abnormal transactions become increasingly covert and technologically sophisticated, financial institutions face significant challenges in analyzing them, particularly in managing the authenticity and validity of customer identities. Unidentified customers are not only a crucial factor in assessing credit and fraud risks but also a key concern in managing abnormal transactions. Therefore, conducting comprehensive and continuous due diligence based on customer risk factors and transaction characteristics to confirm the authenticity and validity of customer identities is a vital risk control measure. However, with a massive customer base, relying solely on manual methods makes it difficult to conduct due diligence and risk control throughout the entire customer lifecycle. Furthermore, limitations in human resources mean that customer due diligence cannot cover the majority of customers. Currently, the industry lacks effective and cost-efficient customer due diligence methods. Summary of the Invention

[0003] This application provides a due diligence report generation method, apparatus, electronic device, and storage medium, which can automate the entire process of large-scale customer due diligence. The technical solution is as follows: According to one aspect of this application, a method for generating a due diligence report is provided, the method comprising: Retrieve object transaction data from multiple candidate objects; Based on a preset identification strategy, the object transaction data is identified, and at least one target object is selected from the multiple candidate objects. Determine the target due diligence script template that matches each of the at least one target object; For each target object, an outbound telephone call is initiated to the target object based on the target due diligence dialogue template to obtain the due diligence voice dialogue of the target object; Based on the due diligence voice dialogue, a target due diligence report for the target object is generated.

[0004] According to another aspect of this application, a due diligence report generation apparatus is provided, the apparatus comprising: The first acquisition module is used to acquire object transaction data of multiple candidate objects; The object recognition module is used to perform object recognition on the object transaction data based on a preset recognition strategy, and to select at least one target object from the plurality of candidate objects; A determination module is used to determine a target due diligence script template that matches each of the at least one target object; The outbound telephone module is used to initiate outbound telephone calls to each target object based on the target due diligence dialogue template, and obtain the due diligence voice dialogue of the target object. The report generation module is used to generate a target due diligence report for the target object based on the due diligence voice dialogue.

[0005] According to one aspect of this application, an electronic device is provided, comprising: a processor and a memory storing a program, the program including instructions that, when executed by the processor, cause the processor to perform the due diligence report generation method as described above.

[0006] According to another aspect of this application, a non-transitory computer-readable storage medium is provided storing computer instructions for causing the computer to perform the due diligence report generation method as described above.

[0007] According to another aspect of this application, a computer program product is provided, comprising computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the aforementioned due diligence report generation method.

[0008] The beneficial effects of the technical solutions provided in this application include at least the following: By automatically screening target objects with abnormal transaction behavior, determining the target due diligence script template for the target object, automatically initiating outbound calls to the target object based on the target due diligence script template, and automatically generating due diligence reports based on the due diligence voice objects obtained from the outbound calls, the entire process of automated customer due diligence is achieved, reducing the reliance on manual processes in the customer due diligence process and improving the efficiency and comprehensiveness of customer due diligence coverage. Attached Figure Description

[0009] Further details, features, and advantages of this application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1 A flowchart of a due diligence report generation method according to an exemplary embodiment of this application is shown; Figure 2 This is a flowchart of a training method for a due diligence model provided in an exemplary embodiment of this application; Figure 3 This is a system architecture diagram of the customer due diligence system provided in this application; Figure 4 This is a schematic diagram of the structure of a due diligence report generation device provided in an embodiment of this application; Figure 5A structural block diagram of an exemplary electronic device that can be used to implement embodiments of this application is shown. Detailed Implementation

[0010] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0011] It should be understood that the steps described in the method embodiments of this application may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this application is not limited in this respect.

[0012] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc., mentioned in this application are only used to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies. It should be noted that the modifications "a" and "a plurality" mentioned in this application are illustrative and not restrictive, and those skilled in the art should understand that unless explicitly indicated in the context, they should be understood as "one or more". The names of messages or information exchanged between multiple devices in the embodiments of this application are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0013] The present invention will now be described with reference to the accompanying drawings. The technical solutions provided by the embodiments of the present invention will be explained in detail through specific examples and application scenarios.

[0014] To reduce the reliance on manual operations in current customer due diligence in the area of ​​abnormal transactions, this application embodiment constructs an automated customer due diligence tool. Through the methods configured in this tool, it is possible to automatically screen abnormal customers, automatically initiate outbound calls to these customers based on corresponding script templates, and automatically generate due diligence reports based on the voice dialogue text obtained from the outbound calls. In other words, the entire customer due diligence process can be automated. Please refer to... Figure 1The diagram illustrates a flowchart of a due diligence report generation method according to an exemplary embodiment of this application. The method is described using an example of its application to an electronic device. Figure 1 As shown, the method includes: Step 101: Obtain object transaction data for multiple candidate objects; Step 102: Based on a preset identification strategy, perform object identification on the object transaction data and select at least one target object from multiple candidate objects; Step 103: Determine the target due diligence script template that matches each of the at least one target object; Step 104: For each target object, make an outbound telephone call to the target object based on the target due diligence dialogue template to obtain the due diligence voice dialogue of the target object; Step 105: Generate a target due diligence report for the target object based on the due diligence voice dialogue.

[0015] For customer due diligence targeting abnormal transactions, the investigation focuses on users who may have engaged in such transactions. Therefore, to minimize disruption to legitimate users, it is necessary to automatically filter out potential users from a massive user base. One possible implementation involves collecting transaction data from multiple candidate users. This allows the candidate users to select one or more target users (or individuals) for whom customer due diligence is required based on this data.

[0016] Users who may exhibit abnormal transactions may have significantly unusual transaction amounts or number of transactions, or their transaction amounts or frequency may be inconsistent with their basic information such as occupation and education. Therefore, to more accurately filter out target users with abnormal transactions from the candidate pool, it is necessary to obtain transaction information such as the transaction amount, number of transactions, and recipient of the transactions, as well as basic information such as the candidate's age, education, occupation, and home or work address, as their corresponding transaction object data. It should be noted that transaction object data must be obtained with the user's authorization.

[0017] Optionally, the transaction data of the candidate can be obtained through the financial or payment apps they use.

[0018] Optionally, when selecting multiple candidates, a T+1 batch investigation strategy can be adopted. That is, based on the current time or the current task execution cycle, batch customer due diligence on the candidates is postponed by one time unit. For example, if the customer due diligence cycle is one week, then customer due diligence on the candidates who have had transaction behavior during that week will be conducted on the day after the end of the week.

[0019] Optionally, the customer due diligence cycle can be a fixed cycle, such as conducting customer due diligence once a week; or it can be a dynamic cycle. If it is a dynamic cycle, the customer due diligence cycle can be dynamically determined by the number of accumulated candidates. For example, a number threshold can be set, and when the number of candidates reaches the number threshold, target object screening and subsequent customer due diligence can be carried out on the accumulated multiple candidates.

[0020] After obtaining the object transaction data of multiple candidate objects, a preset identification strategy (risk user identification strategy) is pre-built. The preset identification strategy can be used to identify multiple candidate objects based on the object transaction data, so as to identify one or more target objects that may have abnormal transaction behavior from multiple candidate objects.

[0021] Optionally, the preset identification strategy includes multiple candidate identification strategies, such as a first candidate identification strategy composed of clustering algorithms, and multiple risk feature identification strategies set by human experience (i.e., second candidate identification strategies). To ensure the accuracy of identifying target objects with abnormal transaction behavior, it is necessary to identify the target transaction data using multiple candidate identification strategies, so as to select at least one target object from multiple candidate objects, and to determine the target hitting strategy for identifying the target object from multiple candidate identification strategies, that is, to determine the candidate identification strategy that the target object is hit by.

[0022] For example, the preset identification strategies include candidate identification strategy A, candidate identification strategy B, candidate identification strategy C, and candidate identification strategy D. Using these strategies respectively, object identification is performed on the transaction data of 10 candidate objects. The identification results are as follows: Object 1, Object 2, and Object 4 are identified as target objects using candidate identification strategy A; Object 5 is identified as a target object using candidate identification strategy B; Object 7 is identified as a target object using candidate identification strategy C; and Object 10 is identified as a target object using candidate identification strategy D. Therefore, Objects 3, 6, 8, and 9 are normal objects and do not require subsequent customer due diligence. Furthermore, the target hit strategy for Objects 1, 2, and 4 is candidate identification strategy A; the target hit strategy for Object 5 is candidate identification strategy B; the target hit strategy for Object 7 is candidate identification strategy C; and the target hit strategy for Object 10 is candidate identification strategy D.

[0023] Taking a pre-defined identification strategy (clustering algorithm) as the first candidate identification strategy as an example, the specific identification process for the target object is as follows: Multiple candidate objects are clustered based on transaction data using the first candidate identification strategy. Outliers in the clustering results are identified as at least one target object. In other words, clustering multiple candidate objects based on transaction data can group objects with similar transaction behaviors into a cluster. Since normal transaction behavior is generally the majority, candidate objects that can be clustered into a cluster are generally users with normal transaction behavior. Outliers (i.e., those significantly different from the cluster) are most likely users with abnormal transaction behavior. Correspondingly, clustering can accurately locate outliers and identify them as target objects that may exhibit abnormal transaction behavior.

[0024] Optionally, to further improve the accuracy of clustering, multiple candidate objects can first be divided into multiple categories based on basic information, and then the candidate objects in each category can be clustered to obtain the outliers (target objects) in each category. When dividing into multiple categories, the division can be based on the basic information of the candidate objects, such as by occupation, by age group, by education level, etc.

[0025] For example, if there are 1,000 candidate objects, and users of the same occupation type generally have similar transaction behaviors, then the 1,000 candidate objects can be divided into 10 categories according to occupation type. Each occupation category can include 100 candidate objects. Then, the 100 candidate objects in each occupation category are clustered. Based on the clustering results, the outliers in that occupation category are determined. By analogy, the outliers corresponding to the 10 occupation categories can be obtained. The sum of these outliers is the target object that may have abnormal transaction behavior among the 1,000 candidate objects.

[0026] Taking the second candidate identification strategy (multiple risk feature identification strategies set by human experience) as an example, the specific identification process of the target object is as follows: Based on the object's transaction data, the second candidate identification strategy performs feature identification, and candidate objects that meet the features are identified as at least one target object. Since the second candidate identification strategy presets multiple risk features of abnormal transaction behaviors, the object's transaction data can be matched with the risk features of abnormal transaction behaviors indicated by the second candidate identification strategy. If the match is successful, the candidate object corresponding to the object's transaction data is identified as the target object; if the match is unsuccessful, the candidate object corresponding to the object's transaction data is a normal user.

[0027] For example, the second candidate identification strategy could be: whether the identity authentication has expired, or whether the number of transactions in a short period of time exceeds a preset threshold.

[0028] Optionally, the second candidate identification strategy can be dynamically updated based on historical experience.

[0029] After identifying potential targets with unusual transaction behavior, customer due diligence needs to be conducted. The typical method for customer due diligence is outbound telephone calls. To reduce reliance on human intervention in customer due diligence, this application also pre-configures multiple candidate due diligence script templates. By determining the target due diligence script template that matches each target, the intelligent voice outbound calling system can automatically initiate outbound telephone calls to the target based on the target due diligence script, thereby improving the efficiency of customer due diligence.

[0030] Considering that different abnormal transaction behaviors may require different wording to thoroughly verify whether they are abnormal users; and that users with different risk levels may also require different wording for analysis and verification, step 103 may also include steps 103A and 103B in an exemplary example.

[0031] Step 103A: For each target object, determine the object risk level of the target object.

[0032] Step 103B: Based on the object's risk level and the target target's corresponding target hit strategy, determine the target due diligence script template that matches the target object from the candidate due diligence script templates.

[0033] The target risk level indicates the degree of credit risk or fraud risk posed by the target entity. A higher target risk level indicates a greater risk of credit risk or fraud, while a lower target risk level indicates a lower risk of credit risk or fraud. The target risk level is determined in advance based on a comprehensive assessment of the target entity's attributes. For example, the target risk level might be automatically generated after the target entity completes a pre-set survey when registering for a financial app (or applying for a credit card).

[0034] Considering that different candidate identification strategies target users with different abnormal transactions, different candidate due diligence script templates can be set according to different candidate identification strategies. Furthermore, when setting due diligence script templates based on the risk levels of different targets, different candidate due diligence script templates will also correspond to different risk levels for the same candidate identification strategy.

[0035] For example, the relationship between candidate dialogue templates, candidate identification strategies, and target risk levels can be shown in Table 1: Table 1

[0036] Table 1 can be pre-configured in electronic devices, so that during the customer due diligence process, after determining the target risk level and target hit strategy for each target object, the target due diligence script template that matches the target object can be determined from the candidate due diligence script templates based on the target risk level and target hit strategy, combined with the correspondence shown in Table 1.

[0037] For example, if the target object's risk level is 2, and the target object is identified through candidate identification strategy B, that is, the target object's target hit strategy is candidate identification strategy B, then according to the relationship shown in Table 1, candidate due diligence script template 5 can be determined as the target due diligence script template for the target object.

[0038] Once the target due diligence script templates for each target object are determined, multiple customer due diligence tasks can be generated and outbound call requests can be sent to the intelligent voice outbound call system. These outbound call requests include customer due diligence tasks for multiple target objects. After receiving the outbound call requests, the intelligent voice outbound call system can initiate batch outbound calls to the target objects on weekdays based on the target due diligence script templates for each target object, complete the due diligence on the target objects, and obtain the due diligence voice dialogues from the target objects.

[0039] In order to ensure that outbound calls reach the target audience and avoid invalid calls, one possible implementation is to initiate outbound calls to the target audience based on a target due diligence script template and obtain the target audience's due diligence voice dialogue. The implementation also includes: determining the target due diligence time for the target audience; initiating outbound calls to the target audience according to the target due diligence time and the target due diligence script template and obtaining the target audience's due diligence voice dialogue.

[0040] To determine the target due diligence time for a target object, one possible implementation is to obtain the target object's historical outbound call status information (historical outbound call time + whether the call was reached), and determine the target due diligence time based on the historical outbound call times when the calls were reached. For example, if the historical outbound call status information indicates that multiple outbound calls made between 9:00 AM and 10:00 AM were not answered (not reached), while multiple outbound calls made between 5:00 PM and 6:00 PM were all answered, then 5:00 PM to 6:00 PM can be determined as the target due diligence time for that target object.

[0041] Optionally, if the target object does not have historical outbound call status information, the historical outbound call status information of other candidate objects that match the basic information of the target object (e.g., occupation) can be selected, and the target due diligence time of the target object can be determined based on the historical outbound call time reached in the historical outbound call status information.

[0042] Optionally, after determining the target due diligence time for the target object, and after making outbound calls according to the target due diligence time can reach the target object, the target due diligence time can be associated with the target object and stored so that the associated target due diligence time can be directly obtained for outbound calls during the next customer due diligence.

[0043] Optionally, if outbound calls are made to the target object according to the target due diligence period, and the target object does not answer (i.e., the outbound call did not reach the target object), in order to avoid frequently disturbing the target object, an outbound call interval can be set. After the outbound call interval is reached, another outbound call can be made to the target object. For example, the outbound call interval can be 2 hours.

[0044] Optionally, if the target object is not reached after the number of repeated outbound calls reaches the threshold, in order to avoid wasting resources, the automatic outbound calls to the target object can be stopped, an unreached flag can be set for the target object, and the object transaction data of the target object can be pushed to the business personnel for subsequent processing.

[0045] Optionally, the device can collect the due diligence reach status of several target objects (i.e., target due diligence time + whether due diligence was successful), and then dynamically adjust the batch outbound call time of subsequent target objects based on the due diligence reach status.

[0046] During outbound calls, the intelligent voice outbound calling system records the due diligence voice conversation with the target and sends it to the intelligent analysis module. The intelligent analysis module converts the voice conversation into due diligence text, which is then used to generate a target due diligence report for the target.

[0047] In an exemplary example, step 105 may also include steps 105A to 105C.

[0048] Step 105A: Convert the due diligence voice dialogue of the target object into due diligence voice text; Step 105B: Obtain the total due diligence frequency of the target object; Step 105C: Input the due diligence voice text, total due diligence frequency, and target transaction data into the trained due diligence model to generate a target due diligence report for the target target.

[0049] To facilitate the analysis of due diligence audio conversations, speech-to-text technology is first used to process the conversations, converting them into text-based due diligence reports. Considering that the target entity's historical due diligence activity and transaction data also influence the final due diligence analysis results, the total due diligence frequency and transaction data are also obtained. These data, along with the audio-based due diligence reports, total due diligence frequency, and transaction data, are then input into a trained due diligence model (AI model). The model analyzes this data to generate a target due diligence report for the target entity.

[0050] For example, a target due diligence report may include: whether the target entity has any abnormal transaction behavior, specific abnormal transaction behavior, credit (transaction) risk analysis of the target entity, etc.

[0051] To ensure the trained due diligence model can generate target due diligence reports, pre-training is required before conducting client due diligence. Please refer to [link / reference needed]. Figure 2 This is a flowchart illustrating a training method for a due diligence model provided in an exemplary embodiment of this application. Figure 2 As shown, the method includes: Step 201: Obtain multiple historical training data pairs. Each historical training data pair includes historical due diligence text, historical due diligence frequency, historical due diligence report, and historical transaction data of the historical object. Step 202: Input the historical due diligence text, historical due diligence frequency and historical transaction data into the initial due diligence model to obtain the predicted due diligence report generated by the initial due diligence model; Step 203: Based on the differences between the predicted due diligence report and the historical due diligence report, train the initial due diligence model, and then iterate through multiple rounds of training to obtain the fully trained due diligence model.

[0052] The model's training data includes multiple historical training data pairs. Each historical training data pair includes historical due diligence text, historical due diligence frequency, historical due diligence reports, and historical transaction data of the historical object. The historical due diligence reports are collected manually analyzed reports.

[0053] In each round of model training, historical due diligence texts, frequencies, and transaction data from historical training data pairs are input into the initial due diligence model to generate a predicted due diligence report. The model's loss function is then determined by comparing the differences between the predicted and historical due diligence reports to train the initial due diligence model. This allows the model to gradually learn how to analyze and generate due diligence reports. After multiple rounds of iterative training, a fully trained due diligence model is obtained after reaching a certain number of iterations; alternatively, a fully trained due diligence model is obtained when the model loss falls below a threshold.

[0054] After the due diligence model that has been trained generates the target due diligence report, the due diligence voice text of the target object and the target due diligence report can be returned to the front end for display to business personnel. Business personnel can then score or revise the target due diligence report based on the due diligence voice text to obtain a revised due diligence report.

[0055] In order to further improve the model performance during the application of the model, revised due diligence reports of each target object can be collected, and the trained due diligence model can be fine-tuned based on the target due diligence reports and the revised due diligence reports.

[0056] Based on the above methods for generating due diligence reports, the system structure of the customer due diligence system (tool) provided in this application can be as follows: Figure 3 As shown. This customer due diligence system may include a configuration module (policy configuration + script configuration), a task management module, an intelligent analysis module, a voice outbound calling system, and a database based on a distributed computing framework.

[0057] The strategy configuration in the configuration module uses clustering algorithms and other methods combined with expert experience to set customer due diligence strategies (preset identification strategies); the script configuration mainly combines expert experience to pre-set robot due diligence script templates for different customer risk levels and strategy differences.

[0058] The task management module, based on the due diligence strategy, sets up scheduled tasks to calculate the characteristic indicators of all customers, generates pending customer due diligence tasks and pushes them out, while simultaneously sending requests to the intelligent voice outbound calling system.

[0059] After receiving a task request from the intelligent customer due diligence system, the intelligent voice outbound calling system makes batch outbound calls on weekdays according to the target customers and the matching scripts to complete the due diligence voice dialogue.

[0060] The intelligent analysis module receives the voice dialogue returned by the voice outbound calling system, converts it into text and stores it as data, and simultaneously activates the AI ​​due diligence assistant. The AI ​​due diligence assistant uses generative AI technology to train a customer due diligence model using historically accumulated customer due diligence texts. The intelligent analysis module returns the target customer's voice dialogue text and the AI ​​assistant's due diligence opinions based on the dialogue text to the task management module, and simultaneously terminates the target customer's due diligence task.

[0061] The due diligence strategy is implemented across the organization's entire customer base and employs multi-layered data cleaning techniques, including data cleaning, deduplication, missing value imputation, and outlier handling, to ensure data quality and accuracy. This provides a reliable data foundation for subsequent data analysis and modeling, reducing analytical errors caused by data quality issues. The data retrieval and execution program is technically abstracted and layered, consisting of rule sets, rules, and rule execution programs. The overall architecture adopts a front-end and back-end interaction model, enabling rapid and automatic mapping and transformation between rule sets and rules. Rule compilation and execution are automated, offering high flexibility. The underlying computation utilizes the Greenplum (a distributed high-performance database) distributed computing framework to achieve high availability and large-scale parallel processing of inherent data retrieval rule operations, thereby quickly completing data extraction.

[0062] In summary, the embodiments of this application provide an automated due diligence method: by automatically screening target objects with abnormal transaction behavior, determining the target due diligence script template for the target objects, automatically initiating outbound calls to the target objects based on the target due diligence script template, and automatically generating due diligence reports based on the due diligence voice objects obtained from the outbound calls, the method achieves full-link automated customer due diligence, reduces manual reliance in the customer due diligence process, and improves the efficiency and comprehensiveness of customer due diligence coverage.

[0063] Please refer to Figure 4 This is a schematic diagram of a due diligence report generation device provided in an embodiment of this application. For example, as shown... Figure 4 As shown, the device 400 includes: The first acquisition module 401 is used to acquire object transaction data of multiple candidate objects; The object recognition module 402 is used to perform object recognition on the object transaction data based on a preset recognition strategy, and to select at least one target object from the plurality of candidate objects; The determining module 403 is used to determine the target due diligence script template that matches each of the at least one target object; The outbound telephone module 404 is used to initiate an outbound telephone call to each target object based on the target due diligence dialogue template, and obtain the due diligence voice dialogue of the target object. The report generation module 405 is used to generate a target due diligence report for the target object based on the due diligence voice dialogue.

[0064] Optionally, the preset identification strategy includes multiple candidate identification strategies; The object recognition module 402 is further configured to: The object transaction data is identified by the multiple candidate identification strategies, at least one target object is selected from the multiple candidate objects, and a target hit strategy for identifying the target object is determined from the multiple candidate identification strategies.

[0065] Optionally, the determining module 403 is further configured to: For each target object, determine the object risk level of that target object; Based on the risk level of the object and the target hit strategy corresponding to the target object, the target due diligence script template that matches the target object is determined from the candidate due diligence script templates.

[0066] Optionally, the object recognition module 402 is further configured to: The first candidate identification strategy is used to cluster the multiple candidate objects based on the object transaction data, and the outliers in the clustering results are identified as at least one target object. The second candidate identification strategy is used to identify features based on the object transaction data, and candidate objects that meet the features are identified as at least one target object.

[0067] Optionally, the report generation module 405 is further configured to: Convert the due diligence voice dialogue of the target object into due diligence voice text; Obtain the total due diligence frequency of the target object; The due diligence voice text, the total due diligence frequency, and the target transaction data are input into the trained due diligence model to generate the target due diligence report for the target object.

[0068] Optionally, the device further includes: The second acquisition module is used to acquire multiple historical training data pairs, each of which includes historical due diligence text, historical due diligence frequency, historical due diligence report and historical transaction data of the historical object. The model prediction module is used to input the historical due diligence text, the historical due diligence frequency and the historical transaction data into the initial due diligence model to obtain the predicted due diligence report generated by the initial due diligence model. The training module is used to train the initial due diligence model based on the difference between the predicted due diligence report and the historical due diligence report, and to obtain the trained due diligence model through multiple rounds of iterative training.

[0069] Optionally, the outbound telephone module 404 is further configured to: Determine the target due diligence time for the target object; According to the target due diligence time and the target due diligence dialogue template, an outbound telephone call is made to the target object to obtain the due diligence voice dialogue of the target object.

[0070] In summary, the embodiments of this application provide an automated due diligence method: by automatically screening target objects with abnormal transaction behavior, determining the target due diligence script template for the target objects, automatically initiating outbound calls to the target objects based on the target due diligence script template, and automatically generating due diligence reports based on the due diligence voice objects obtained from the outbound calls, the method achieves full-link automated customer due diligence, reduces manual reliance in the customer due diligence process, and improves the efficiency and comprehensiveness of customer due diligence coverage.

[0071] An exemplary embodiment of this application also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor. The memory stores a computer program executable by the at least one processor, the computer program, when executed by the at least one processor, causing the electronic device to perform a due diligence report generation method according to an embodiment of this application.

[0072] An exemplary embodiment of this application also provides a non-transitory computer-readable storage medium storing a computer program, wherein the computer program, when executed by a computer's processor, is used to cause the computer to perform a due diligence report generation method according to an embodiment of this application.

[0073] An exemplary embodiment of this application also provides a computer program product, including a computer program, wherein, when executed by a computer's processor, the computer program is used to cause the computer to perform a due diligence report generation method according to an embodiment of this application.

[0074] refer to Figure 5 The present invention describes a structural block diagram of an electronic device 500 that can serve as a server or client of this application, which is an example of a hardware device that can be applied to various aspects of this application. The electronic device is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the application described and / or claimed herein.

[0075] like Figure 5As shown, the electronic device 500 includes a computing unit 501, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 508 into a random access memory (RAM) 503. The RAM 503 may also store various programs and data required for the operation of the electronic device 500. The computing unit 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0076] Multiple components in electronic device 500 are connected to I / O interface 505, including: input unit 506, output unit 507, storage unit 508, and communication unit 509. Input unit 506 can be any type of device capable of inputting information to electronic device 500. Input unit 506 can receive input digital or character information and generate key signal inputs related to user settings and / or function control of electronic device. Output unit 507 can be any type of device capable of presenting information and may include, but is not limited to, a display, speaker, video / audio output terminal, vibrator, and / or printer. Storage unit 508 may include, but is not limited to, disk and optical disk. Communication unit 509 allows electronic device 500 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and / or chipsets, such as Bluetooth devices, WiFi devices, WiMax devices, cellular communication devices, and / or the like.

[0077] The computing unit 501 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 501 performs the various methods and processes described above. For example, in some embodiments, Figure 1 , Figure 2 The method shown can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 500 via ROM 502 and / or communication unit 509. In some embodiments, computing unit 501 can be configured to execute by any other suitable means (e.g., by means of firmware). Figure 1 , Figure 2 The method shown.

[0078] The program code used to implement the methods of this application may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0079] In the context of this application, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0080] As used in this application, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, device, and / or apparatus (e.g., disk, optical disk, memory, programmable logic device (PLD)) for providing machine instructions and / or data to a programmable processor, including machine-readable media that receive machine instructions as machine-readable signals. The term "machine-readable signal" refers to any signal for providing machine instructions and / or data to a programmable processor.

[0081] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0082] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0083] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

Claims

1. A due diligence report generation method characterized by, The method comprises: acquiring object transaction data of a plurality of candidate objects; performing object identification on the object transaction data based on a preset identification strategy, and screening at least one target object from the plurality of candidate objects; determining a target due diligence dialogue template matched with each target object in the at least one target object; for each target object, initiating a telephone outbound call to the target object based on the target due diligence dialogue template, and obtaining a due diligence voice dialogue of the target object; generating a target due diligence report of the target object based on the due diligence voice dialogue.

2. The method of claim 1, wherein, The preset identification strategy comprises a plurality of candidate identification strategies; The object identification on the object transaction data based on the preset identification strategy, and screening at least one target object from the plurality of candidate objects, comprises: performing object identification on the object transaction data by the plurality of candidate identification strategies respectively, screening at least one target object from the plurality of candidate objects, and determining a target hit strategy for identifying the target object from the plurality of candidate identification strategies.

3. The method of claim 2, wherein, The determination of the target due diligence dialogue template matched with each target object in the at least one target object comprises: for each target object, determining an object risk level of the target object; based on the object risk level and the target hit strategy corresponding to the target object, determining the target due diligence dialogue template matched with the target object from candidate due diligence dialogue templates.

4. The method of claim 2, wherein, The object identification on the object transaction data by the plurality of candidate identification strategies respectively, and screening at least one target object from the plurality of candidate objects, comprises: performing clustering on the plurality of candidate objects based on the object transaction data by a first candidate identification strategy, and determining an outlier object in a clustering result as the at least one target object; performing feature identification based on the object transaction data by a second candidate identification strategy, and determining a candidate object meeting a feature as the at least one target object.

5. The method according to any one of claims 1 to 4, characterized in that, The generation of the target due diligence report of the target object based on the due diligence voice dialogue comprises: converting the due diligence voice dialogue of the target object into due diligence voice text; acquiring a total due diligence frequency of the target object; inputting the due diligence voice text, the total due diligence frequency, and the object transaction data into a trained due diligence model, and generating the target due diligence report of the target object.

6. The method of claim 5, wherein, The method further comprises: acquiring a plurality of historical training data pairs, each historical training data pair comprising historical due diligence text, historical due diligence frequency, historical due diligence report, and historical transaction data of a historical object; inputting the historical due diligence text, the historical due diligence frequency, and the historical transaction data into an initial due diligence model, and obtaining a predicted due diligence report generated by the initial due diligence model; based on a difference between the predicted due diligence report and the historical due diligence report, training the initial due diligence model to obtain the trained due diligence model through multiple rounds of iterative training.

7. The method according to any one of claims 1 to 4, characterized in that, The initiation of the telephone outbound call to the target object based on the target due diligence dialogue template, and the obtaining of the due diligence voice dialogue of the target object, comprise: determine a target follow-up time of the target object; initiate a phone outbound call to the target object according to the target follow-up time and the target follow-up script template, and obtain a follow-up voice conversation of the target object.

8. A due diligence report generation apparatus characterized by comprising: The device comprises: a first acquisition module configured to acquire object transaction data of a plurality of candidate objects; an object identification module configured to perform object identification on the object transaction data based on a preset identification strategy, and filter at least one target object from the plurality of candidate objects; a determination module configured to determine a target follow-up script template matched with each target object in the at least one target object; a phone outbound call module configured to initiate a phone outbound call to each target object based on the target follow-up script template, and obtain a follow-up voice conversation of the target object; a report generation module configured to generate a target follow-up report of the target object based on the follow-up voice conversation.

9. An electronic device comprising: a processor; and a memory storing a program, wherein the program comprises instructions that, when executed by the processor, cause the processor to perform the follow-up report generation method according to any one of claims 1-7.

10. A non-transitory computer readable storage medium having stored thereon computer instructions, wherein, The computer instructions are used to cause the computer to perform the follow-up report generation method according to any one of claims 1-7.

Images