Simple and efficient case division implementation method
Through the combination of preset case classification rules and case classification data, combined with the matching degree and the highest expected value algorithm for single allocation, the scalability and maintenance difficulty of case classification in the existing technology is solved, and efficient, flexible allocation and resources of cases are achieved. Optimize configuration.
Patent Information
- Application Number
- CN202411010778.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-26
- Publication Date
- 2024-11-12
AI Technical Summary
The existing technology lacks scalability and maintenance difficulty in case classification, and the newly added classification dimensions are difficult to integrate into existing rules, resulting in complex case classification strategies and difficult to achieve optimal allocation of resources.
Different case division strategies are formed by combining preset case division rules and case division data, and cases are allocated to the most matching organizations and staff using the matching degree algorithm and a single-time allocation highest expectation algorithm to be used to allocate cases to the most matched organizations and staff, realizing a dynamic and scalable case division engine.
It realizes simple and efficient allocation of cases, the algorithm is flexible and scalable, and the diversity of case-dividing data is seamlessly connected with the algorithm, improving the efficiency of resource allocation and the flexibility of case-dividing strategies.
Smart Images

Figure CN118941035A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and in particular to a simple and efficient case division implementation method. Background Art
[0002] Case division is an important task in the entire post-loan processing field. Its essence is to let the most suitable person handle the most suitable case at the most suitable time with the most suitable effort, so as to achieve the optimal allocation of resources. The work of handling case division is generally carried out from two aspects. On the one hand, different processing agencies are formed in the organization, such as: internal processing agencies and external processing agencies. Different teams are also planned within the organization, such as the M3+ (overdue for more than 3 months) team and the on-site processing team, so that more professional people can do more difficult things. On the other hand, in terms of processing tools, it is necessary to establish an intelligent processing system to undertake efficient case division strategies.
[0003] The case division strategy includes many dimensions, such as: number of cases, number of days overdue / age of cases, overdue amount, outstanding principal, amount required for early settlement, case region, staff rating, agency / team / individual processing performance, and some label attributes of cases, such as: whether to borrow new to repay old, whether it is the first overdue, etc. The more dimensions involved in case division, the more complex the implementation of the case division strategy. To this end, it is necessary to combine the big data platform and the label platform to collect various data to establish a case division data model, and combine it with the rule matching algorithm to realize a dynamic, scalable, configurable, and low-cost operation and maintenance case division engine.
[0004] In the actual implementation process, independent branch development is mostly carried out according to various rules, and the case division algorithm and data model are not independent, which makes the case division strategy non-scalable and difficult to maintain. If a new case division dimension is added, it is difficult to integrate it into the existing case division rules. Summary of the invention
[0005] The purpose of the present invention is to provide a simple and efficient method for implementing case division in view of the deficiencies in the prior art.
[0006] To achieve the above objectives, the present invention provides a simple and efficient case division implementation method, comprising:
[0007] Different case division strategies are formed according to preset case division rules and a combination of information in case division data, and corresponding routing rules are formulated according to different case division strategies, so that the cases to be assigned execute the corresponding case division strategies through the routing rules, and the case division data includes information about the case, information about the organization receiving the case, and information about the staff of the organization;
[0008] Acquire current case division data, select a case division strategy for the current case division data, and extract corresponding information from the current case division data according to the selected case division strategy;
[0009] Calculate the matching degree between the case and the organization receiving the case based on the matching degree algorithm and the extracted information, and assign the case to the organization with the highest matching degree;
[0010] The case is assigned to the employee in the organization with the highest matching degree according to the single-assignment highest expected value algorithm.
[0011] Furthermore, the case information includes loan note information and employee information. The loan note information includes the number of cases, overdue days, account age, outstanding principal, overdue amount, early settlement amount and new loan to repay old debt. The employee information includes place of origin, product, whether lost contact and score card.
[0012] Furthermore, the organization used to receive cases includes a processing team and an outsourced agency; information about the processing team includes processing products, field visits, and an M3+ team; information about the outsourced agency includes the number of employees, collection rate, location, and field visits.
[0013] Furthermore, the preset case distribution rules include average distribution, proportion distribution, regional distribution, distribution by collection rate and distribution by collection rate.
[0014] Furthermore, the method of allocating the case to the employee of the organization with the highest matching degree according to the single allocation maximum expected value algorithm is as follows:
[0015] Calculate the expected value of a single allocation for each employee;
[0016] Sort the employees from smallest to largest according to their expected single allocation value;
[0017] The case closest to the expected value of a single allocation among the unallocated cases is allocated to the corresponding employee, and the expected value of a single allocation of the employee is recalculated. If the recalculated expected value of a single allocation of the employee is less than or equal to 0, the employee is removed from the employees to be allocated. Otherwise, it is determined whether the number of employees to be allocated is 0. If the number of employees to be allocated is not 0, a circular allocation is performed, otherwise the allocation is terminated.
[0018] Beneficial effects: 1. The present invention realizes the allocation of cases through the matching algorithm and the single allocation maximum expected value algorithm. The two algorithms are relatively simple and easy to implement. The two algorithms have different focuses. The matching algorithm matches the case to the organization, and then allocates the case to a specific employee within the organization through the single allocation maximum expected value algorithm.
[0019] 2. The algorithm is flexible and scalable;
[0020] 3. Diversity of case data, and seamless connection with the algorithm through configuration. By setting the weight attribute, the role of the attribute can be increased in the matching algorithm. By configuring the attribute of the attribution algorithm, it can be specified in which algorithm the attribute is valid;
[0021] 4. Since the case classification data and case classification algorithm are separated, the data module is responsible for the collection and calculation of case classification data. The data can come from the big data platform, labeling platform, external system, or the staff can label the cases during the processing process. These data in turn affect the case classification. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 It is a schematic diagram of a simple and efficient case division implementation method of an embodiment of the present invention;
[0023] Figure 2 It is a flowchart of allocating cases based on the single-assignment maximum expected value algorithm. DETAILED DESCRIPTION
[0024] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. The embodiments are implemented based on the technical solution of the present invention. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention.
[0025] like Figure 1 and Figure 2 As shown, the embodiment of the present invention provides a simple and efficient method for implementing case division, including:
[0026] Different case division strategies are formed according to the preset case division rules and several information combinations in the case division data, and corresponding routing rules are formulated according to different case division strategies, so that the cases to be assigned can execute the corresponding case division strategies through the routing rules. The case division data includes the information of the case, the information of the organization used to receive the case, and the information of the staff under the above organization. The above preset case division rules include average distribution, proportion distribution, regional distribution, distribution by collection rate, and distribution by collection rate. The case information includes loan information and staff information. The loan information includes the number of cases, overdue days, account age, outstanding principal, overdue amount, early settlement amount, and new loan repayment. The staff information includes native place, product, whether lost contact, and score card. The organization used to receive the case includes the processing team and the outsourcing agency. The information of the processing team includes the processing product, outbound visit, and M3+ team. The information of the outsourcing agency includes the number of staff, collection rate, location, and outbound visit. The case division data provides the basis for the subsequent algorithms. Each type of data needs to have specific quantitative indicators, as shown in Table 1:
[0027] Table 1:
[0028]
[0029]
[0030] Different case division strategies can be used in combination, as shown in Table 2:
[0031] Table 2:
[0032]
[0033]
[0034] The current case division data is obtained, a case division strategy is selected for the current case division data, and corresponding information is extracted from the current case division data according to the selected case division strategy.
[0035] The matching degree between the case and the organization receiving the case is calculated based on the matching degree algorithm and the extracted information, and the case is assigned to the organization with the highest matching degree. Specifically, the matching degree algorithm can be expressed as: matching degree = ∑ (matching data × data weight). An example is shown in Table 3:
[0036] Table 3:
[0037]
[0038] Here, we assume that the weight of the outsourced data is 1.5, and the other data are all 1. Then, the matching degree formula calculates that the outsourced agency C has the highest matching degree, and the case is assigned to the outsourced agency C. If there are the same matching degrees, it can be simply handled by random allocation.
[0039] Assign the case to the employee in the organization with the highest matching score based on the highest expected value of a single assignment algorithm. Figure 2 , the implementation is as follows:
[0040] Calculate the expected value of a single allocation for each employee.
[0041] Employees are sorted from smallest to largest according to their expected value of single allocation.
[0042] The case closest to the expected value of a single allocation among the unallocated cases is allocated to the corresponding employee, and the expected value of a single allocation of the employee is recalculated. If the recalculated expected value of a single allocation of the employee is less than or equal to 0, the employee is removed from the employees to be allocated. Otherwise, it is determined whether the number of employees to be allocated is 0. If the number of employees to be allocated is not 0, a circular allocation is performed, otherwise the allocation is terminated.
[0043] Taking the “proportion allocation + number of cases + overdue amount” case allocation strategy, 10 cases, and 3 employees assigned to the above-mentioned outsourcing agency C as an example, the current case allocation data is shown in Table 4:
[0044] Table 4:
[0045] case Overdue amount Case 1 50 Case 2 150 Case 3 120 Case 4 200 Case 5 10 Case 6 50 Case 7 170 Case 8 200 Case 9 40 Case 10 10
[0046] The three employees account for 10%, 20% and 70% respectively. The specific processing logic is as follows:
[0047] 1. Calculate the total number of cases and the total amount of cases expected by the staff for this allocation
[0048] Expected number of cases for staff = total number of cases × proportion - number of cases assigned;
[0049] Expected amount of case assigned by an employee = (total case amount × proportion - total amount of assigned cases) / number of cases expected by the employee
[0050] Before the first allocation, the number of allocated cases and the total amount are both 0. The calculation results are as follows:
[0051] a Staff member's expected number of cases = 1
[0052] Expected amount of a staff member's single case = 100
[0053] bThe number of cases expected by the staff = 2
[0054] b Employee's expected amount for a single case = 100
[0055] cThe number of cases expected by the staff = 7
[0056] c Employee's expected amount for a single case = 100
[0057] After calculating the expected value of employee allocation, the employees are sorted from small to large according to the expected number of cases: a, b, c.
[0058] 2. Loop through the employee set and start assigning cases. Take the case amount closest to the employee's expected amount of cases assigned in a single assignment from the unassigned cases. In the above calculation, the expected amount of single assignments for the three employees is 100. Among the 10 unassigned cases, case 3 is the closest to 100. Assign case 3 to employee a. If the expected value of a changes, recalculation is required. When the employee's expected number of cases or the expected amount of single assignment is less than or equal to 0, it means that the employee has been assigned and needs to be removed from the set of employees to be assigned.
[0059] The expected value changes of this example allocation process are shown in Table 5:
[0060] Table 5:
[0061]
[0062] In summary, the present invention has the following advantages:
[0063] 1. The present invention realizes the allocation of cases through a matching algorithm and a single allocation maximum expected value algorithm. The two algorithms are relatively simple and easy to implement. The two algorithms have different focuses. The matching algorithm matches the case to the organization, and then allocates the case to a specific employee within the organization through a single allocation maximum expected value algorithm.
[0064] 2. The algorithm is flexible and scalable. For example, when allocating cases to a batch of processing agencies based on average distribution + number of cases + overdue amount + overdue days, since there are three groups of case data, the processing agencies have three expected values. Therefore, when allocating cases, the overdue amount and overdue days of all cases are calculated and subtracted from the expected value, and the two comparison results are added, that is, |number of overdue days for cases-expected number of days for a single allocation|+|overdue amount for cases-expected amount for a single allocation|, and the smallest value is taken and allocated to the current agency.
[0065] 3. Diversity of case data, and seamless connection with the algorithm through configuration. By setting the weight attribute, the role of this attribute can be increased in the matching algorithm. By configuring the attribute of the attribute, you can specify in which algorithm the attribute is effective. The specific examples are shown in Table 6:
[0066] Table 6:
[0067]
[0068] 4. Since the case division data and the case division algorithm are separated, the data module is responsible for the collection and calculation of case division data. The data can come from the big data platform, label platform, external system, or the staff can label the case during the processing process. These data in turn affect the case division. For example, if the staff cannot contact the customer through various means, the customer will be labeled as lost contact. The lost contact label and the external visit label of the processing agency can be matched. In this way, the matching algorithm will be used to assign the case to the on-site processing team the next time the case is divided. The entire process is automatically driven by the system.
[0069] In order to test the effect of the present invention, the collection rates of different account ages were statistically analyzed and compared before and after optimization. The comparison data is shown in Table 7, and the collection rate has been significantly improved.
[0070] Table 7:
[0071] Aging Not optimized After optimization M1-M3 35% 52% M3+ 23% 33%
[0072] The above is only a preferred embodiment of the present invention. It should be noted that for ordinary technicians in this technical field, other parts not specifically described belong to the prior art or common knowledge. Several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention.
Claims
1. A simple and efficient method for implementing case division, characterized in that: include: Different case division strategies are formed according to preset case division rules and a combination of information in case division data, and corresponding routing rules are formulated according to different case division strategies, so that the cases to be assigned execute the corresponding case division strategies through the routing rules, and the case division data includes information about the case, information about the organization receiving the case, and information about the staff of the organization; Acquire current case division data, select a case division strategy for the current case division data, and extract corresponding information from the current case division data according to the selected case division strategy; Calculate the matching degree between the case and the organization receiving the case based on the matching degree algorithm and the extracted information, and assign the case to the organization with the highest matching degree; The case is assigned to the employee in the organization with the highest matching degree according to the single-assignment highest expected value algorithm.
2. A simple and efficient case division implementation method according to claim 1, characterized in that: The case information includes loan note information and employee information. The loan note information includes the number of cases, overdue days, account age, outstanding principal, overdue amount, early settlement amount and new loan to repay old debt. The employee information includes place of origin, product, whether lost contact and score card.
3. A simple and efficient case division implementation method according to claim 1, characterized in that: The organization used to receive cases includes the processing team and the outsourcing agency; the information of the processing team includes the processing products, outsourcing visits and the M3+ team; the information of the outsourcing agency includes the number of employees, collection rate, location and outsourcing visits.
4. A simple and efficient case division implementation method according to claim 1, characterized in that: The preset case distribution rules include average distribution, proportion distribution, regional distribution, distribution by collection rate and distribution by collection rate.
5. A simple and efficient case division implementation method according to claim 1, characterized in that: The method of allocating the case to the employee of the organization with the highest matching degree according to the single allocation maximum expected value algorithm is as follows: Calculate the expected value of a single allocation for each employee; Sort the employees from smallest to largest according to their expected single allocation value; The case closest to the expected value of a single allocation among the unallocated cases is allocated to the corresponding employee, and the expected value of a single allocation of the employee is recalculated. If the recalculated expected value of a single allocation of the employee is less than or equal to 0, the employee is removed from the employees to be allocated. Otherwise, it is determined whether the number of employees to be allocated is 0. If the number of employees to be allocated is not 0, a circular allocation is performed, otherwise the allocation is terminated.
