Overdue case collection method and device, storage medium and computer equipment

Abstract

In the overdue case collection method and device, the storage medium and the computer equipment provided by the invention, in response to the overdue case collection request, the case data of each overdue case in the overdue case collection request is processed through the plurality of nodes in the SpringBatch, and the target classification data of each overdue case is obtained; the target classification data comprises first-level classification data and second-level classification data; based on the first-level classification data of each overdue case, determining a first-level queue of each overdue case, and based on the second-level classification data of each overdue case, determining a second-level queue corresponding to each overdue case in each first-level queue; and determining an initial collection strategy corresponding to each secondary queue, and performing content adjustment on the initial collection strategy of the secondary queue where each overdue case is located according to the target classification data of each overdue case to obtain a target collection means of each overdue case. The differentiation processing can improve the collection success rate, thereby improving the overall rate of return.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, device, storage medium and computer equipment for collecting overdue cases. Background Art

[0002] Financial debt collection is an important part of handling non-performing assets in lending business. When borrowers fail to repay on time, the debt collection department or outsourced debt collection company will collect the debt through phone calls, text messages, emails, door-to-door visits, etc. to ensure the return of funds to financial institutions and reduce overdue losses. With the development of the financial market, the phenomenon of overdue payments has increased, and financial debt collection has become an indispensable part of maintaining the healthy operation of financial institutions.

[0003] At present, the main method of debt collection is still manual telephone collection, supplemented by digital means such as text messages and emails. In some special cases, debt collectors will communicate with the debtor at home, or even sue for debt recovery through legal means. Some financial institutions have also tried to improve efficiency and reduce costs by introducing intelligent debt collection systems, such as automated calls and voice robots. However, the overall debt collection method still relies on relatively traditional and single means, and cannot make differentiated treatments based on the characteristics of different cases, resulting in low debt collection efficiency. Summary of the invention

[0004] The purpose of this application is to solve at least one of the above-mentioned technical defects, especially the overall collection method in the prior art still relies on relatively traditional and single means, and is unable to make differentiated treatments according to the characteristics of different cases, thus resulting in a technical defect of low collection efficiency.

[0005] In a first aspect, the present application provides a method for collecting overdue cases, the method comprising:

[0006] In response to the overdue case collection request, the case data of each overdue case in the overdue case collection request is processed through multiple nodes in SpringBatch to obtain target classification data of each overdue case, where the target classification data includes first-level classification data and second-level classification data;

[0007] Based on the first-level classification data of each overdue case, determine the first-level queue corresponding to each overdue case, and based on the second-level classification data of each overdue case, determine the second-level queue corresponding to each overdue case in each first-level queue;

[0008] Determine the initial collection strategy corresponding to each secondary queue, and adjust the content of the initial collection strategy of the secondary queue where each overdue case is located based on the target classification data of each overdue case to obtain the target collection means for each overdue case.

[0009] In one embodiment, the step of processing case data of each overdue case through multiple nodes in SpringBatch to obtain target classification data of each overdue case includes:

[0010] In SpringBatch, trigger any node to read the case data of each overdue case, obtain the task corresponding to each overdue case, and send each task to the target topic channel through the message queue;

[0011] When each task is sent successfully, each node is triggered to subscribe to the target topic channel, so that each node uses the cursor method to batch load and process each task according to the target topic channel to obtain the target classification data of each overdue case.

[0012] In one embodiment, the method further comprises:

[0013] When any node reads the case data of each overdue case, the case data of each overdue case is loaded into the memory and the task timer of the node is started;

[0014] Determine the amount of remaining tasks in memory based on the number of tasks processed by each node;

[0015] If the remaining task amount is not 0, and the timing duration of the task timer exceeds the preset timing duration, the database where the case data of each overdue case is located is queried to determine the case data processing progress of each overdue case.

[0016] In one embodiment, based on the first-level classification data of each overdue case, the step of determining the first-level queue corresponding to each overdue case includes:

[0017] According to the first classification data of each overdue case, the risk level of each overdue case is determined, and the overdue cases are grouped according to the risk level of each overdue case to obtain a first-level queue corresponding to each overdue case.

[0018] In one embodiment, the step of determining the secondary queue corresponding to each overdue case in each primary queue based on the secondary classification data of each overdue case includes:

[0019] For each first-level queue, the overdue cases in the first-level queue are grouped according to the risk level and the second classification data of each overdue case in the first-level queue to obtain a second-level queue corresponding to each overdue case in the first-level queue.

[0020] In one embodiment, the step of determining an initial collection strategy corresponding to each secondary queue includes:

[0021] For each secondary queue, split the secondary queue into a first test queue and a second test queue, obtain the first test collection strategy preset for the first test queue, and obtain the second test collection strategy preset for the second test queue, and based on the preset test cycle, determine the collection rates corresponding to the first test collection strategy and the second test collection strategy, respectively, and determine the initial collection strategy corresponding to the secondary queue in the first test collection strategy and the second test collection strategy according to each collection rate.

[0022] In one embodiment, the steps of adjusting the content of the initial collection strategy of the secondary queue where each overdue case is located according to the target classification data of each overdue case to obtain the target collection means for each overdue case include:

[0023] Determine the severity of collection for each overdue case based on the target classification data of each overdue case;

[0024] According to the severity of collection for each overdue case and the initial collection strategy of the secondary queue, the collection content for each overdue case is determined to obtain the target collection method for each overdue case.

[0025] In a second aspect, the present application provides a device for collecting overdue cases, the device comprising:

[0026] A target classification data determination module is used to respond to the overdue case collection request, process the case data of each overdue case in the overdue case collection request through multiple nodes in SpringBatch, and obtain the target classification data of each overdue case, the target classification data including the first-level classification data and the second-level classification data;

[0027] A queue determination module, for determining a first-level queue corresponding to each overdue case based on the first-level classification data of each overdue case, and determining a second-level queue corresponding to each overdue case in each first-level queue based on the second-level classification data of each overdue case;

[0028] The target collection means determination module is used to determine the initial collection strategy corresponding to each secondary queue, and adjust the content of the initial collection strategy of the secondary queue where each overdue case is located according to the target classification data of each overdue case, so as to obtain the target collection means for each overdue case.

[0029] In a third aspect, the present application provides a storage medium: the storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the overdue case collection method in any of the above embodiments.

[0030] In a fourth aspect, the present application provides a computer device, comprising: one or more processors, and a memory;

[0031] Computer-readable instructions are stored in the memory. When the computer-readable instructions are executed by one or more processors, the steps of the overdue case collection method in any of the above embodiments are performed.

[0032] It can be seen from the above technical solutions that the embodiments of the present application have the following advantages:

[0033] In the overdue case collection method, device, storage medium and computer equipment provided in the present application, firstly, by utilizing multiple nodes in Spring Batch to process the data of overdue cases, large-scale cases can be efficiently processed in parallel, which greatly improves the processing speed and data computing ability, not only saving collection time, but also improving resource utilization, reducing reliance on manual intervention, and reducing operating costs; secondly, the method performs detailed processing through the target classification data of the case, including first-level classification data and second-level classification data. Each overdue case determines a first-level queue based on the first-level classification data, and is further subdivided into a second-level queue based on the second-level classification data, so that different cases can be assigned to corresponding queues according to their characteristics, so as to implement targeted collection strategies according to the different characteristics of the cases, thereby improving the efficiency and pertinence of the collection; in addition, the method flexibly adjusts the initial collection strategy in combination with the classification data of the case to ensure that the collection means of each case are in line with its specific circumstances. Through this flexible collection strategy, the collection plan can be tailored according to the characteristics of different types of cases, avoiding the disadvantages of the single traditional collection means and lack of differentiation. Such differentiated processing can improve the success rate of debt collection, thereby increasing the overall collection rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0035] Figure 1 A flowchart of a method for collecting overdue cases provided in an embodiment of the present application;

[0036] Figure 2 A schematic diagram of the structure of the overdue case collection device provided in an embodiment of the present application;

[0037] Figure 3 A schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0039] The present application provides a method for collecting overdue cases. The following embodiments are described by taking the method applied to a computer device as an example. It can be understood that the computer device can be any device with data processing functions, including but not limited to a single server, a server cluster, a personal laptop computer, a desktop computer, etc. Figure 1 As shown, the method may include the following steps:

[0040] S101: In response to a collection request for overdue cases, the case data of each overdue case in the collection request for overdue cases is processed through multiple nodes in SpringBatch to obtain target classification data for each overdue case, where the target classification data includes first-level classification data and second-level classification data.

[0041] Among them, the overdue case collection request refers to the collection request initiated for those cases where the debt is not repaid within the specified period. Spring Batch is a Java-based framework, mainly used for large-scale data processing, capable of processing batch tasks, and providing support for multi-threading, multi-step processing, retry and fault recovery. A node can be a server, computer or virtual machine. Each node is responsible for processing a part of the work tasks. Multiple nodes mean that the collection request of the case will be processed in parallel by multiple computing nodes, increasing the processing capacity and speed. Case data refers to the relevant information of each overdue case, including which investors are overdue for the case, the number of overdue days, the age of the account, the risk level, the case data, the overdue amount, the settlement amount, the penalty interest, the queue grouping, the collection type, etc. The target classification data is the case data after processing will be classified into different categories so that the corresponding collection strategy can be adopted. The target classification data can contain two levels of classification. The first level of classification data is the preliminary classification of the case, such as the classification according to the overdue duration or amount. The second level of classification data is to further subdivide the case based on the first level of classification, which may be based on more complex rules, such as the debtor's previous repayment behavior or debt priority.

[0042] In this step, the collection request for each overdue case can be submitted to SpringBatch in the form of a batch task. Spring Batch will decompose the task into multiple steps, such as data reading, processing, and writing, according to the configured workflow, and assign them to different nodes for parallel processing. Each node can execute a specific classification rule engine to classify the case data by step. When processing case data, the classification algorithm or the preset rule engine is used to calculate the target classification based on the case characteristics.

[0043] It can be understood that by processing case data in parallel through Spring Batch, the computing power of multiple nodes can be fully utilized to improve the efficiency of data processing. Due to the large amount of case data and complex processing, single-node processing is not only slow, but also easily causes excessive system load. Parallel processing ensures load balancing and reduces data processing failures caused by individual node failures through the automatic retry mechanism of batch tasks. Grading and classification can adopt more refined collection strategies based on different case characteristics. For example, the first-level classification can quickly conduct preliminary screening of cases and give priority to the most urgent or most valuable cases. The second-level classification can select collection methods more specifically, improve the collection success rate, and reduce resource waste.

[0044] S102: Based on the first-level classification data of each overdue case, determine the first-level queue corresponding to each overdue case, and based on the second-level classification data of each overdue case, determine the second-level queue corresponding to each overdue case in each first-level queue.

[0045] Among them, the first-level queue is a collection that contains cases belonging to the same first-level classification. Each first-level queue represents a major case category, which is convenient for centralized processing and management. The second-level queue contains cases belonging to the same second-level classification. Each second-level queue is more finely classified according to the characteristics of the cases in the first-level queue, which is convenient for implementing specific collection strategies.

[0046] In this step, by processing the first-level classification data of each overdue case and assigning it to the corresponding first-level queue, all case data can be traversed, and the classification information of each case can be mapped to a pre-defined queue according to the classification rules, such as using hash mapping, conditional judgment, etc. After completing the classification of the first-level queue, based on the second-level classification data of each overdue case, the cases in each first-level queue are further processed, and the cases in the first-level queue can be traversed again, and the cases can be divided into the corresponding second-level queues according to the second-level classification standards, such as the debtor's credit rating, the urgency of the case, etc. The division of the second-level queue can also use similar hash mapping or conditional logic.

[0047] It can be understood that efficient case management can be achieved by dividing overdue cases into different queues according to the first-level classification data. The division of the first-level queue allows the processing process to proceed in an orderly manner, so that the processing strategy for the same type of cases can be unified, which helps to improve work efficiency and reduce management complexity. In addition, the collection strategy for a specific queue can be more targeted to ensure the rational allocation of resources. Further subdividing the first-level queue into second-level queues can make the collection strategy more precise. The cases corresponding to each second-level queue are more similar in characteristics, allowing the appropriate collection method or strategy to be selected according to the specific circumstances of the case. This refinement can improve the collection success rate and reduce unnecessary collection costs.

[0048] S103: Determine the initial collection strategy corresponding to each secondary queue, and adjust the content of the initial collection strategy of the secondary queue where each overdue case is located according to the target classification data of each overdue case, so as to obtain the target collection means for each overdue case.

[0049] Among them, the initial collection strategy refers to the preliminary collection plan set for a specific secondary queue, which may include telephone collection, SMS reminders, email notifications, etc. These strategies are based on the overall analysis of the cases in the secondary queue and aim to improve the collection success rate. The target collection method is a specific adjustment result of the initial collection strategy, aiming to develop a more targeted collection method based on the characteristics of each overdue case. These methods can be an optimization based on the initial strategy to ensure that the handling of each case can achieve the best results.

[0050] In this step, the case characteristics of each secondary queue are read from the definition of the secondary queue, and the corresponding initial collection strategy is assigned to each queue. For example, a predefined rule or strategy library can be used to match the case characteristics in the secondary queue to select the most suitable collection strategy. Based on the target classification data of each overdue case, the characteristics of the case can be analyzed and the content of the initial collection strategy in the corresponding secondary queue can be adjusted. For example, a feature matching algorithm is used in combination with a machine learning model to evaluate the specific circumstances of each case, and the content of the initial collection strategy is adjusted based on the results to obtain the best collection method.

[0051] It is understandable that by setting the initial collection strategy for each secondary queue, it is possible to quickly respond to the processing needs of different types of cases, making the collection process more targeted and improving processing efficiency. Adjusting the collection strategy based on the target classification data of each overdue case can significantly improve the collection success rate. The characteristics of each overdue case may be different, and flexible strategy adjustments ensure that the most appropriate method can be used for collection, thereby improving the overall debt recovery rate and reducing collection costs.

[0052] Furthermore, the overdue case collection method provided in this application can be verified through traffic playback. Specifically, the decision engine is deployed separately as another microservice to provide decision-making services. Traffic playback means that the system captures the recent online decision engine interface traffic input parameters and responses from time to time for saving. When the decision engine makes relevant engine package adjustments or code changes, these captured and saved input parameters will automatically call the service to be launched, and visually compare the differences in the output to ensure the correctness of this modification. The specific workflow is as follows:

[0053] A: The decision engine service captures the recent production environment traffic input and response and saves them;

[0054] B: This modified version is released to the pre-release environment;

[0055] C: Enable playback, and the latest online traffic input will call the modified version of the pre-release environment;

[0056] D: After the playback is completed, the similarities and differences of the response parameters can be visually compared.

[0057] If the relevant policies of the decision engine have not been changed, then in theory the response of each traffic should be consistent with that of the production environment. If the relevant policies of the decision engine have been changed, it is necessary to check whether the scope and value of this change are in line with expectations.

[0058] In the above embodiment, firstly, by using multiple nodes in Spring Batch to process the data of overdue cases, large-scale cases can be processed in parallel efficiently, which greatly improves the processing speed and data computing ability, not only saves the collection time, but also improves the resource utilization, reduces the reliance on manual intervention, and reduces the operating cost; secondly, the method refines the target classification data of the case, including the first-level classification data and the second-level classification data. Each overdue case determines the first-level queue based on the first-level classification data, and further subdivides into the second-level queue based on the second-level classification data, so that different cases can be assigned to the corresponding queues according to their characteristics, so as to implement targeted collection strategies according to the different characteristics of the cases, thereby improving the efficiency and pertinence of the collection; in addition, the method flexibly adjusts the initial collection strategy in combination with the classification data of the case to ensure that the collection means of each case are in line with its specific circumstances. Through this flexible collection strategy, the collection plan can be tailored according to the characteristics of different types of cases, avoiding the disadvantages of the singleness and lack of differentiation of traditional collection means. Such differentiated processing can improve the success rate of collection, thereby improving the overall collection rate.

[0059] In one embodiment, the step of processing case data of each overdue case through multiple nodes in SpringBatch to obtain target classification data of each overdue case includes:

[0060] In SpringBatch, trigger any node to read the case data of each overdue case, obtain the task corresponding to each overdue case, and send each task to the target topic channel through the message queue;

[0061] When each task is sent successfully, each node is triggered to subscribe to the target topic channel, so that each node uses the cursor method to batch load and process each task according to the target topic channel to obtain the target classification data of each overdue case.

[0062] Among them, a task is a definition of a specific operation or work unit, which in this embodiment refers to the processing work performed on each overdue case. A message queue is a technology for asynchronous communication, through which messages can be passed between different systems or components. In this embodiment, a message queue is used to transmit processing requests for each task. A target topic channel is a concept in a message queue that is used to identify a category of a group of messages. All related task messages are sent to this channel for consumption by subscribers. In data processing, a cursor is a method for traversing a data set. In this embodiment, the cursor method refers to each node gradually reading and processing tasks in a message queue when processing tasks.

[0063] Specifically, an event-driven mechanism can be set up. When the trigger conditions are met, such as time intervals and changes in system status, any node will start the task of reading overdue cases. This can be configured through Spring Batch steps. The node connects to the database or other data sources and uses the corresponding reader to load the case data. After reading the case data, each node will package it into a task message and send the task to the specified target topic channel through the producer interface of the message queue. Each node will trigger the subscription to the target topic channel in the callback of the successful sending of the task. After subscribing, the node will use the consumer interface of the message queue to read the messages in the target topic in cursor mode, process each task and generate the target classification data.

[0064] In one example, the data processing in the SpringBatch task node overview is non-streaming. The entire processing flow is divided into N multiple dependent processing nodes. After one node is processed, the next node is processed. Each individual task node has three independent functional areas: data extraction, data processing, and data input. It is a streaming processing process. The cursor method is used to load data in batches during data extraction, such as querying the database cursor method to return a single data. After the data processing is completed, it is temporarily stored in the memory buffer. After accumulating a certain amount or a certain period of time, a batch persistence is performed. In this way, the memory of the application can be effectively controlled to be relatively stable, and the memory will not be blown up due to massive data. It is also friendly to the database, without repeated and frequent calls, reducing DB pressure and network transmission time. Furthermore, the retry feature of springBatch itself can be fully utilized to retry when a single batch of data is processed abnormally. Among them, the exception refers to the specified exception thrown in the business code during task processing, or an unknown exception. Only specified exceptions or unknown exceptions will be retried. Exceptions that are not important to the main process can be ignored, tolerated, and skipped. However, if the exception cannot be ignored or skipped and the retry is clear and cannot succeed, the task will be stopped and manually investigated and resolved through alarms. If the overall task is restarted, the successfully executed task nodes can also be skipped to quickly restore to the position of the pending nodes before the restart.

[0065] In this embodiment, triggering the node to read case data can ensure flexible and efficient response to new data processing requests, avoid idle resources, and ensure that task processing can be carried out at the appropriate time, thereby improving overall processing efficiency. Sending tasks through message queues can reduce the coupling between task sending and processing, allowing each node to work independently. In addition, the use of message queues can also provide task persistence. Even in the event of a failure, unprocessed tasks can still be consumed again to ensure data security. Batch loading tasks through cursors can efficiently process large amounts of data, reading gradually rather than loading all at once, which can effectively manage memory usage and increase data processing speed. In addition, batch processing can reduce the number of network transmissions, reduce latency, improve overall system performance, and ensure that the target classification data for each overdue case can be generated in a timely and accurate manner.

[0066] In one embodiment, the method further comprises:

[0067] When any node reads the case data of each overdue case, the case data of each overdue case is loaded into the memory and the task timer of the node is started;

[0068] Determine the amount of remaining tasks in memory based on the number of tasks processed by each node;

[0069] If the remaining task amount is not 0, and the timing duration of the task timer exceeds the preset timing duration, the database where the case data of each overdue case is located is queried to determine the case data processing progress of each overdue case.

[0070] Among them, memory is the short-term storage space of the computer. When a node executes a task, it will load the case data into the memory so that the data can be quickly accessed and processed. Task timer is a tool used to track the execution time of a task. When the node starts processing a case, it starts the timer to help monitor the progress of task execution. The remaining task volume refers to the number of tasks that have not been processed in the memory. By calculating the tasks that have been completed by each node, the remaining task volume can be determined. The preset timing duration refers to the expected processing time limit set for the task. When the task exceeds this time, further inspection is required. The processing progress indicates the current execution status of a task and is used to determine the degree of completion of the task in the entire processing process.

[0071] Specifically, when any node reads the case data, it can load the case data from the database or other data sources into the memory, and then start a timer. The timer function can be used to record the execution time of the task, or the listener monitoring mechanism of springBatch can be used to easily expand and monitor the execution of each task node, such as printing logs or writing to the database, visualizing the progress, etc. By monitoring the number of tasks completed by the node, the remaining tasks in the memory are updated in real time. For example, this can be achieved through a simple counter or a task queue management system based on the memory state. For example, each time a task is completed, the counter is reduced by 1 until the remaining tasks are 0.

[0072] When the remaining task amount is not 0 and the timing duration exceeds the preset timing duration, the current status of the task can be obtained by querying the processing progress field in the database. For example, this can be achieved through SQL statements or using data access layer technologies such as Spring DataJPA to return the processing progress of each case data.

[0073] In one example, the master node distributes tasks, aggregates tasks, and the child nodes perform task calculations. For example, a certain node A (machine A) is the master node, and other machines (machine B, machine C, machine D) are child nodes. Assume that a single node takes 2 hours to process 100 million data. If the four nodes A, B, C, and D all participate in the calculation, the processing can be completed in 0.5 hours under ideal conditions. The specific processing details are as follows:

[0074] a: The master node calculates the total number of tasks and writes it to Redis, and starts a local timer to periodically check whether the tasks are completed;

[0075] b: The master node distributes tasks, retrieves data and sends it to a specific Topic via MQ;

[0076] c: Each child node subscribes to this Topic (the master node itself can also participate), and after the data is processed, it is atomically decremented in Redis;

[0077] d: When the local timer of the master node checks that the amount of completed data is equal to the total number of tasks, or if it is not strongly dependent on Redis and the time consumed is greater than the configurable experience value, there is a query DB to determine whether it is completed. It is determined that this task node has been processed and the next task node is processed.

[0078] In this embodiment, loading case data into memory can greatly improve the processing speed, because the memory access speed is much higher than reading data from a hard disk or a database. By starting the task timer, the execution progress of the task can be effectively monitored. When the execution time of a task exceeds expectations, the timer will trigger the corresponding subsequent operations to prevent the task from falling into infinite waiting or processing timeout. If the task processing time is too long, querying the task processing progress in the database helps to identify and resolve potential delays or failures, prevent tasks from being stuck for a long time, and ensure the timeliness and reliability of task processing.

[0079] In one embodiment, based on the first-level classification data of each overdue case, the step of determining the first-level queue corresponding to each overdue case includes:

[0080] According to the first classification data of each overdue case, the risk level of each overdue case is determined, and the overdue cases are grouped according to the risk level of each overdue case to obtain a first-level queue corresponding to each overdue case.

[0081] Among them, the risk level is the case risk category assessed based on the first classification data. The risk level is usually divided into multiple levels, such as low risk, medium risk, and high risk, which are used to measure the priority and urgency of each overdue case.

[0082] Specifically, based on the first classification data of each overdue case, the risk level of the case is determined by a preset rule or algorithm, which can be achieved through conditional judgment statements, rule engines (such as Drools) or machine learning models. For example, cases can be divided into different risk levels based on factors such as the number of overdue days and the amount of debt owed. Once the risk level of the case is determined, all overdue cases can be grouped according to the risk level.

[0083] In this embodiment, the determination of risk levels can handle overdue cases in an orderly manner. Through grading, different processing strategies and resources can be assigned to cases of different risk levels, thereby optimizing resource allocation, reducing inefficient repetitive work, and improving the success rate of collection. Through grouping, similar types of cases can be processed in batches more efficiently. The generation of a first-level queue can perform targeted operations on cases of different risk levels, avoiding the chaos of disorderly processing. Setting different processing strategies and priorities for each queue can improve the efficiency of the entire collection process.

[0084] In one embodiment, based on the second-level classification data of each overdue case, the step of determining the second-level queue corresponding to each overdue case in each first-level queue includes:

[0085] For each first-level queue, the overdue cases in the first-level queue are grouped according to the risk level and the second classification data of each overdue case in the first-level queue to obtain a second-level queue corresponding to each overdue case in the first-level queue.

[0086] Specifically, the overdue cases in each primary queue are grouped again. First, the risk level of each overdue case is checked and further refined using the secondary classification data. For example, similar types of cases can be grouped based on fields such as the number of overdue days and the debtor's credit score. Based on the grouping results, a corresponding secondary queue is generated for each case.

[0087] In this embodiment, further grouping is performed based on the first-level queue, which enables more accurate case handling. Although the first-level queue has been preliminarily grouped according to risk level, different cases may still have significant differences within the same risk level. By introducing the second classification data, these cases can be more finely distinguished, and a more appropriate handling method can be determined based on specific conditions, ensuring efficient and accurate management of each case and improving the collection success rate.

[0088] In one embodiment, the step of determining the initial collection strategy corresponding to each secondary queue includes:

[0089] For each secondary queue, split the secondary queue into a first test queue and a second test queue, obtain the first test collection strategy preset for the first test queue, and obtain the second test collection strategy preset for the second test queue, and based on the preset test cycle, determine the collection rates corresponding to the first test collection strategy and the second test collection strategy, respectively, and determine the initial collection strategy corresponding to the secondary queue in the first test collection strategy and the second test collection strategy according to each collection rate.

[0090] Among them, the first test queue and the second test queue are two sub-queues divided into two secondary queues, which are used to apply different collection strategies respectively. The first test collection strategy and the second test collection strategy are two different preset collection strategies, which are applied to the first test queue and the second test queue respectively. The preset test cycle is a time period for testing the effectiveness of the collection strategy. During this cycle, the system will track and record the performance of each strategy, such as the collection rate. The collection rate is the ratio of successfully recovered funds through a specific collection strategy within the preset test cycle. The collection rate is an important indicator for evaluating the effectiveness of the collection strategy.

[0091] Specifically, the overdue cases in the secondary queue can be split into two sub-queues according to the preset splitting rules or random number generation mechanism. For example, the system can use a random allocation algorithm to assign 50% of the cases to the first test queue and the remaining 50% to the second test queue. Each test queue will have a corresponding collection strategy. The two collection strategies can be loaded from the preset strategy library and applied to the corresponding test queues. After the preset test cycle is over, the collection rate is calculated based on the collection results of each test queue. The calculation formula for the collection rate is usually the amount of money successfully recovered divided by the total amount owed. Based on the collection rate of each collection strategy, the effects of the two strategies can be compared, and the strategy with a higher collection rate can be selected as the initial collection strategy for the secondary queue.

[0092] In one example, cases with similar characteristics are in the same secondary queue, and this secondary queue can be split into two groups. For example, if you want to test the effects of arbitration SMS and ordinary SMS in the M2 queue, set a 20-day test period, and when creating an SMS sending strategy, specify two groups to send arbitration SMS and ordinary SMS respectively, and compare the collection rates every certain period.

[0093] In this embodiment, by splitting the secondary queue into two test queues, different collection strategies can be A / B tested to effectively evaluate the performance of different strategies on similar cases, thereby providing data support for the selection of global strategies. Splitting the queues also helps avoid the blind application of the same strategy to all cases and improves the flexibility of the collection strategy. By testing different collection strategies, it is possible to find out which strategy is most effective for a specific case type. The collection rate data during the test period provides a basis for optimizing the collection process, which can significantly improve the overall collection efficiency and avoid invalid or inefficient collection behaviors.

[0094] In one embodiment, according to the target classification data of each overdue case, the initial collection strategy of the secondary queue where each overdue case is located is adjusted to obtain the target collection means for each overdue case, including:

[0095] Determine the severity of collection for each overdue case based on the target classification data of each overdue case;

[0096] According to the severity of collection for each overdue case and the initial collection strategy of the secondary queue, the collection content for each overdue case is determined to obtain the target collection method for each overdue case.

[0097] The severity of collection refers to the intensity of collection efforts determined based on the specific circumstances of each overdue case, usually based on factors such as the duration of the arrears, the amount, and the credit status of the debtor. The severity of collection can be divided into multiple levels, ranging from gentle reminders to compulsory measures.

[0098] Specifically, the severity of collection can be determined by analyzing the target classification data of each overdue case and applying preset rules or algorithms. For example, the severity of collection can be calculated using conditional judgment or scoring systems based on data such as the number of days overdue, the debtor's financial situation, and the amount owed. Cases with higher risks may require more stringent collection measures, while low-risk cases may only require gentle reminders. Once the severity of collection is determined, the appropriate collection content is selected based on the severity and the initial collection strategy of the secondary queue to which the case belongs. This can be performed through a mapping table, with different collection actions corresponding to each collection severity level, and the final collection content is adjusted in combination with the queue's initial strategy to obtain the corresponding target collection means.

[0099] In this embodiment, different overdue cases have different situations. Simply adopting a unified collection method may lead to waste of resources or missed collection opportunities. By accurately setting the severity of collection according to the target classification data, it is possible to ensure that each case receives a collection effort that matches its risk and characteristics, maximizing the collection success rate. By combining the severity of collection and the initial strategy of the queue, the most appropriate collection content is selected for each case, which can not only reduce excessive collection of low-risk cases and improve user experience, but also ensure that high-risk cases receive the necessary severity and ensure the success rate of collection, optimize the use of collection resources, and ensure the accuracy and flexibility of the collection process.

[0100] The following describes the overdue case collection device provided in the embodiment of the present application. The overdue case collection device described below and the overdue case collection method described above can be referred to in correspondence with each other. Figure 2 As shown, the present application provides a device for collecting overdue cases, the device comprising:

[0101] The target classification data determination module 201 is used to respond to the overdue case collection request, process the case data of each overdue case in the overdue case collection request through multiple nodes in SpringBatch, and obtain the target classification data of each overdue case, wherein the target classification data includes the first-level classification data and the second-level classification data;

[0102] A queue determination module 202, for determining a first-level queue corresponding to each overdue case based on the first-level classification data of each overdue case, and determining a second-level queue corresponding to each overdue case in each first-level queue based on the second-level classification data of each overdue case;

[0103] The target collection means determination module 203 is used to determine the initial collection strategy corresponding to each secondary queue, and adjust the content of the initial collection strategy of the secondary queue where each overdue case is located according to the target classification data of each overdue case, so as to obtain the target collection means for each overdue case.

[0104] In one embodiment, the target classification data determination module 201 includes:

[0105] The task sending unit is used to trigger any node in SpringBatch to read the case data of each overdue case, obtain the task corresponding to each overdue case, and send each task to the target topic channel through the message queue;

[0106] The target classification data determination unit is used to trigger each node to subscribe to the target subject channel when each task is sent successfully, so that each node can batch load and process each task in a cursor manner according to the target subject channel to obtain the target classification data of each overdue case.

[0107] In one embodiment, the apparatus further comprises:

[0108] A task timer start module is used to load the case data of each overdue case into the memory and start the task timer of the node when any node reads the case data of each overdue case;

[0109] The remaining task quantity determination module is used to determine the remaining task quantity in the memory according to the number of tasks processed by each node;

[0110] The processing progress determination module is used to query the database where the case data of each overdue case is located if the remaining task amount is not 0 and the timing duration of the task timer exceeds the preset timing duration to determine the case data processing progress of each overdue case.

[0111] In one embodiment, the queue determination module 202 includes:

[0112] The first-level queue determination unit is used to determine the risk level of each overdue case according to the first classification data of each overdue case, and group the overdue cases according to the risk level of each overdue case to obtain the first-level queue corresponding to each overdue case.

[0113] In one embodiment, the queue determination module 202 includes:

[0114] The secondary queue determination unit is used to group the overdue cases in each primary queue according to the risk level and second classification data of each overdue case in the primary queue to obtain the secondary queue corresponding to each overdue case in the primary queue.

[0115] In one embodiment, the target collection method determination module 203 includes:

[0116] The initial collection strategy determination unit is used to split each secondary queue into a first test queue and a second test queue, obtain the first test collection strategy preset for the first test queue, and obtain the second test collection strategy preset for the second test queue, and determine the collection rates corresponding to the first test collection strategy and the second test collection strategy respectively based on the preset test cycle, and determine the initial collection strategy corresponding to the secondary queue in the first test collection strategy and the second test collection strategy according to the various collection rates.

[0117] In one embodiment, the target collection method determination module 203 includes:

[0118] A collection severity determination unit, used to determine the collection severity corresponding to each overdue case based on the target classification data of each overdue case;

[0119] The target collection means determination unit is used to determine the collection content corresponding to each overdue case according to the collection severity corresponding to each overdue case and the initial collection strategy of the secondary queue to obtain the target collection means for each overdue case.

[0120] In one embodiment, the present application also provides a storage medium, in which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the overdue case collection method as described in any of the above embodiments.

[0121] In one embodiment, the present application also provides a computer device having computer-readable instructions stored therein. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the overdue case collection method as described in any one of the above embodiments.

[0122] Indicatively, if Figure 3 As shown, Figure 3 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. The computer device 300 may be provided as a server. Figure 3 The computer device 300 includes a processing component 302, which further includes one or more processors, and a memory resource represented by a memory 301 for storing instructions executable by the processing component 302, such as an application. The application stored in the memory 301 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 302 is configured to execute instructions to execute the overdue case collection method of any of the above embodiments.

[0123] The computer device 300 may further include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate based on an operating system stored in the memory 301, such as Windows Server TM, Mac OS X TM, Unix TM, Linux TM, Free BSD TM, or the like.

[0124] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0125] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the term "include", "comprise" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not clearly listed, or also includes elements inherent to such process, method, article or equipment. In the absence of more restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article or equipment including the elements. Herein, "one", "one", "said", "the" and "it" may also include plural forms, unless the context clearly indicates another way. A plurality refers to at least two cases, such as 2, 3, 5 or 8, etc. "And / or" includes any and all combinations of the relevant listed items.

[0126] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can refer to each other.

[0127] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for collecting overdue cases, characterized in that: The method comprises: In response to a collection request for overdue cases, the case data of each overdue case in the collection request for overdue cases is processed through multiple nodes in SpringBatch to obtain target classification data of each overdue case, wherein the target classification data includes first-level classification data and second-level classification data; Based on the first-level classification data of each overdue case, determine the first-level queue corresponding to each overdue case, and based on the second-level classification data of each overdue case, determine the second-level queue corresponding to each overdue case in each first-level queue; Determine the initial collection strategy corresponding to each of the secondary queues, and adjust the content of the initial collection strategy of the secondary queue where each overdue case is located according to the target classification data of each overdue case, so as to obtain the target collection means for each overdue case.

2. The overdue case collection method according to claim 1 is characterized in that: The step of processing the case data of each overdue case through multiple nodes in SpringBatch to obtain target classification data of each overdue case includes: In the SpringBatch, any node is triggered to read the case data of each overdue case, obtain the task corresponding to each overdue case, and send each task to the target topic channel through the message queue; When each of the tasks is sent successfully, each of the nodes is triggered to subscribe to the target subject channel, so that each of the nodes uses a cursor method to batch load and process each of the tasks according to the target subject channel to obtain the target classification data of each overdue case.

3. The overdue case collection method according to claim 2 is characterized in that: The method further comprises: When any node reads the case data of each of the overdue cases, the case data of each of the overdue cases is loaded into the memory, and the task timer of the node is started; Determine the amount of remaining tasks in the memory according to the number of tasks processed by each of the nodes; If the remaining task amount is not 0, and the timing duration of the task timer exceeds the preset timing duration, the database where the case data of each overdue case is located is queried to determine the case data processing progress of each overdue case.

4. The overdue case collection method according to claim 1 is characterized in that: The step of determining the first-level queue corresponding to each overdue case based on the first-level classification data of each overdue case comprises: According to the first classification data of each overdue case, the risk level of each overdue case is determined, and the overdue cases are grouped according to the risk level of each overdue case to obtain a first-level queue corresponding to each overdue case.

5. The overdue case collection method according to claim 4 is characterized in that: The step of determining the secondary queue corresponding to each overdue case in each primary queue based on the second-level classification data of each overdue case comprises: For each of the first-level queues, the overdue cases in the first-level queue are grouped according to the risk level of each overdue case in the first-level queue and the second classification data to obtain a second-level queue corresponding to each overdue case in the first-level queue.

6. The overdue case collection method according to claim 1, characterized in that: The step of determining the initial collection strategy corresponding to each of the secondary queues includes: For each of the secondary queues, the secondary queue is split into a first test queue and a second test queue, the first test collection strategy preset for the first test queue is obtained, and the second test collection strategy preset for the second test queue is obtained, and based on the preset test cycle, the collection rates corresponding to the first test collection strategy and the second test collection strategy are determined respectively, and according to the various collection rates, the initial collection strategy corresponding to the secondary queue is determined in the first test collection strategy and the second test collection strategy.

7. The overdue case collection method according to claim 1 is characterized in that: The step of adjusting the content of the initial collection strategy of the secondary queue where each overdue case is located according to the target classification data of each overdue case to obtain the target collection means for each overdue case includes: Determine the severity of collection corresponding to each overdue case according to the target classification data of each overdue case; According to the severity of collection corresponding to each overdue case and the initial collection strategy of the secondary queue, the collection content corresponding to each overdue case is determined to obtain the target collection means for each overdue case.

8. A device for collecting overdue cases, characterized in that: The device comprises: A target classification data determination module is used to respond to an overdue case collection request, process the case data of each overdue case in the overdue case collection request through multiple nodes in SpringBatch, and obtain target classification data for each overdue case, wherein the target classification data includes first-level classification data and second-level classification data; A queue determination module, configured to determine a first-level queue corresponding to each overdue case based on the first-level classification data of each overdue case, and to determine a second-level queue corresponding to each overdue case in each first-level queue based on the second-level classification data of each overdue case; The target collection means determination module is used to determine the initial collection strategy corresponding to each of the secondary queues, and adjust the content of the initial collection strategy of the secondary queue where each overdue case is located according to the target classification data of each overdue case, so as to obtain the target collection means for each overdue case.

9. A storage medium, characterized in that: The storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the overdue case collection method as described in any one of claims 1 to 7.

10. A computer device, characterized in that: include: one or more processors, and memory; The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the one or more processors, the steps of the overdue case collection method as described in any one of claims 1 to 7 are performed.

Images