Complaint risk determination and prediction model training method and device, equipment and medium

Abstract

The embodiment of the invention discloses a complaint risk determination and prediction model training method and device, equipment and a medium. The complaint risk prediction model training method comprises the steps of obtaining a plurality of historical complaint lists of historical insurance cases; marking the historical insurance cases as positive samples or negative samples according to complaint time and complaint level fields of the historical complaint lists of the historical insurance cases; for historical insurance cases corresponding to the positive samples and the negative samples, determining field weights of complaint fields according to field values of the complaint fields of historical complaint lists of the historical insurance cases and the number of the historical complaint lists, and determining a target complaint field from the complaint fields according to the field weights; and training the complaint risk prediction model according to the field weight of the target complaint field. According to the technical scheme provided by the invention, the complaint risk prediction model for accurately predicting the complaint risk can be trained, and the training accuracy of the complaint risk prediction model is improved.

Description

Technical Field

[0001] The present invention relates to the field of deep learning, and in particular to a training method, device, equipment and medium for complaint risk determination and prediction model. Background Art

[0002] In the insurance field, customers often complain about the quality of business services. Currently, complaints are mainly divided into internal complaints and regulatory complaints. Internal complaints are complaints from customers within the company regarding insurance cases. The scope of impact of complaints is limited. Most internal complaints will be resolved after internal processing. However, there are some internal complaints that customers are not satisfied with the results of the processing and will upgrade the internal complaints to regulatory complaints. Regulatory complaints are high-level complaints against insurance companies, and they will have a wide range of impacts on the reputation and business of insurance companies.

[0003] Currently, we can only judge the customer’s attitude based on the experience of the person in charge of the complaint, and then determine whether the complaint will escalate. However, the current method has low accuracy in judging complaint risks and cannot meet the current risk prediction needs. Summary of the invention

[0004] The technical solution of the present invention provides a method, device, equipment and medium for training a complaint risk determination and prediction model. Through the technical solution of the embodiment of the present invention, a complaint risk prediction model that accurately predicts complaint risks can be trained, thereby improving the accuracy of the training of the complaint risk prediction model.

[0005] In a first aspect, an embodiment of the present invention provides a method for training a complaint risk prediction model, comprising:

[0006] Acquire multiple historical complaint sheets of historical insurance cases, wherein the historical complaint sheets include complaint time and complaint fields, and the complaint fields include complaint level fields;

[0007] According to the complaint time and complaint level fields of the historical complaint sheets of the historical insurance cases, the historical insurance cases are marked as positive samples or negative samples, wherein there is a progressive relationship of complaint levels between the historical complaint sheets corresponding to the positive samples, and the complaint levels represent the severity of the complaints;

[0008] For the historical insurance cases corresponding to the positive samples and the negative samples, determine the field weight of the complaint field according to the field value of the complaint field of the historical complaint sheet of the historical insurance case and the number of historical complaint sheets, and determine the target complaint field from the complaint field according to the field weight;

[0009] The complaint risk prediction model is trained according to the field weight of the target complaint field.

[0010] In a second aspect, an embodiment of the present invention provides a method for determining complaint risk, including:

[0011] Obtaining a current complaint form and a field weight table of a current insurance case, wherein the current complaint form includes complaint fields, and the field weight table is determined according to complaint fields of historical complaint forms and the number of historical complaint forms;

[0012] Determine the target complaint field and the corresponding field weight according to the complaint field and field weight table of the current complaint form;

[0013] Based on the complaint risk prediction model, the complaint escalation probability of the current insurance case is determined according to the field weight of the target complaint field;

[0014] The complaint risk of the current insurance case is determined based on the complaint escalation probability.

[0015] In a third aspect, an embodiment of the present invention provides a training device for a complaint risk prediction model, comprising:

[0016] An acquisition module, used for acquiring a plurality of historical complaint sheets of historical insurance cases, wherein the historical complaint sheets include a complaint time and a complaint field, and the complaint field includes a complaint level field;

[0017] A labeling module, used for labeling the historical insurance cases as positive samples or negative samples according to the complaint time and complaint level fields of the historical complaint sheets of the historical insurance cases, wherein there is a progressive relationship of complaint levels between the historical complaint sheets corresponding to the positive samples, and the complaint levels represent the severity of the complaints;

[0018] A target complaint field determination module is used to determine the field weight of the complaint field for the historical insurance cases corresponding to the positive samples and the negative samples according to the field value of the complaint field of the historical complaint sheets of the historical insurance cases and the number of historical complaint sheets, and determine the target complaint field from the complaint fields according to the field weight;

[0019] A training module is used to train the complaint risk prediction model according to the field weight of the target complaint field.

[0020] In a fourth aspect, an embodiment of the present invention provides a complaint risk determination device, including:

[0021] A case acquisition module, used to acquire a current complaint form and a field weight table of a current insurance case, wherein the current complaint form includes a complaint field, and the field weight table is determined according to the complaint fields of historical complaint forms and the number of historical complaint forms;

[0022] A weight determination module, used to determine the target complaint field and the corresponding field weight according to the complaint field and field weight table of the current complaint form;

[0023] An input module, used to determine the complaint escalation probability of the current insurance case according to the field weight of the target complaint field based on the complaint risk prediction model;

[0024] The risk determination module is used to determine the complaint risk of the current insurance case based on the complaint escalation probability.

[0025] In a fifth aspect, an embodiment of the present invention provides an electronic device, the electronic device comprising:

[0026] at least one processor; and,

[0027] a memory communicatively connected to the at least one processor; wherein,

[0028] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the complaint risk prediction model or the complaint risk determination method described in any one of the embodiments of the present invention.

[0029] In a sixth aspect, an embodiment of the present invention provides a computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable a processor to implement a training method for a complaint risk prediction model or a complaint risk determination method as described in any one of the embodiments of the present invention when executed.

[0030] An embodiment of the present invention provides a training method, device, equipment and medium for a complaint risk determination and prediction model. The training method of the complaint risk prediction model includes: obtaining multiple historical complaint forms of historical insurance cases, wherein the historical complaint forms include complaint time and complaint fields, and the complaint fields include complaint level fields; marking the historical insurance cases as positive samples or negative samples based on the complaint time and complaint level fields of the historical complaint forms of the historical insurance cases, wherein there is a progressive relationship between the complaint levels between the historical complaint forms corresponding to the positive samples, and the complaint levels represent the severity of the complaints; for the historical insurance cases corresponding to the positive samples and the negative samples, determining the field weight of the complaint field based on the field value of the complaint field of the historical complaint forms of the historical insurance cases and the number of historical complaint forms, and determining the target complaint field from the complaint field based on the field weight; training the complaint risk prediction model based on the field weight of the target complaint field. Specifically, by accurately labeling positive samples in historical insurance cases and determining the target complaint field from the complaint field based on the field weight, the accuracy of model training can be improved and the resource consumption of the training process can be reduced. Furthermore, this method can train a complaint risk prediction model that accurately predicts complaint risks, thereby improving the training efficiency of the complaint risk prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 A flowchart of a method for training a complaint risk prediction model provided in Example 1 of the present invention;

[0034] Figure 2 A flowchart of a method for training a complaint risk prediction model provided in Embodiment 2 of the present invention;

[0035] Figure 3 A flowchart of a method for determining complaint risk provided in Embodiment 3 of the present invention;

[0036] Figure 4 A schematic diagram of the structure of a training device for a complaint risk prediction model provided in Embodiment 4 of the present invention;

[0037] Figure 5 A schematic diagram of the structure of a complaint risk determination device provided in Embodiment 5 of the present invention;

[0038] Figure 6 A schematic diagram of the structure of an electronic device provided in Embodiment 6 of the present invention. DETAILED DESCRIPTION

[0039] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0040] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0041] It should be noted that the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in the technical solution of the present disclosure are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0042] Embodiment 1

[0043] Figure 1 A flowchart of a method for training a complaint risk prediction model provided in Example 1 of the present invention. The complaint risk prediction model trained by this method can be used to predict the risk of complaint escalation in the insurance field. The method can be executed by a training device for a complaint risk prediction model. The device can be executed by training the complaint risk prediction model. The device can be composed of software and / or hardware and configured in a computer or server.

[0044] like Figure 1 As shown, including:

[0045] Step 110: Obtain multiple historical complaint forms of historical insurance cases, wherein the historical complaint forms include complaint time and complaint fields, and the complaint fields include a complaint level field.

[0046] Among them, historical insurance cases are insurance cases generated in the historical process, and historical complaint forms are documents generated by customers' complaints about historical insurance cases. The historical complaint forms include complaint time and complaint fields, among which the complaint field includes the complaint level field. The complaint field can also include the historical complaint form number, case number, complaint source channel, province, complaint acceptance time, complaint filing time, complaint closing time, complaint reason, complaint appeal, complaint description, complaint handling results, complaint policy number, insurance type name, policy sales channel, payment type, first payment amount, policy effective time, agent number, whether the agent is employed, sales agency and sales project and other field data.

[0047] Furthermore, the complaint field includes a complaint level field, which is used to characterize the complaint level of the complaint form. For example, the complaint level includes internal complaints and regulatory complaints, wherein internal complaints are complaints made by customers within the company regarding insurance cases. The scope of impact of complaints is limited. Most internal complaints will be resolved after internal processing. However, there are still some internal complaints that customers are not satisfied with the processing results and will upgrade the internal complaints to regulatory complaints. Regulatory complaints are high-level complaints against insurance companies. Regulatory complaints will have a greater impact on the reputation and business of insurance companies. Therefore, internal complaints should be avoided as much as possible from being upgraded to regulatory complaints. The complaint risk prediction model trained by the method of the embodiment of the present invention can accurately determine the probability of complaint upgrade of the insurance case corresponding to the complaint form. For example, if the probability of complaint upgrade is 50%, it indicates that after the current round of internal complaints, there is a 50% probability that the customer will upgrade the internal complaint to a regulatory complaint. Through the complaint risk prediction model, insurance cases with high complaint risks can be accurately detected at the early stage of the complaint, so that staff can intervene in advance to avoid the negative impact caused by the upgraded complaint.

[0048] Step 120: Mark the historical insurance case as a positive sample or a negative sample based on the complaint time and complaint level fields of the historical complaint form of the historical insurance case, wherein there is a progressive relationship of complaint levels between the historical complaint forms corresponding to the positive samples, and the complaint level represents the severity of the complaint.

[0049] Among them, there is a progressive relationship of complaint levels between the historical complaint forms corresponding to the positive samples, and the complaint level represents the severity of the complaint. Specifically, the historical insurance policies of the positive samples include insurance policies with internal complaints and insurance policies with regulatory complaints, and the complaint levels of the insurance policies with internal complaints and those with regulatory complaints are in a progressive relationship, that is, the severity of regulatory complaints is higher than that of internal complaints.

[0050] Specifically, before training the complaint risk prediction model, samples need to be labeled, that is, historical insurance cases need to be determined as positive samples or negative samples based on historical complaint sheets of historical insurance cases.

[0051] Optionally, if there are at least two historical complaint forms for the same historical insurance case, whose complaint level fields indicate a progressive relationship in complaint level, and if there is a complaint time interval between the two historical complaint forms that is less than a preset interval, then the historical insurance case is determined as a positive sample, otherwise it is determined as a negative sample.

[0052] Specifically, the progressive relationship represents that the complaint level for the same insurance case has risen from internal complaints to regulatory complaints. Furthermore, since complaints themselves have a time limit, the time interval between internal complaint sheets with a progressive relationship is less than the preset interval. If it is greater than the preset interval, it means that the two complaints are separate independent complaints and there is no internal correlation. The historical insurance case corresponding to the historical complaint sheet is not the required positive sample. If it is less than the preset interval, it means that the two complaints are relational complaints with an internal connection. The regulatory complaint is caused by the escalation of the internal complaint, and the corresponding historical insurance case is a positive sample.

[0053] Step 130: For the historical insurance cases corresponding to the positive samples and negative samples, determine the field weight of the complaint field based on the field value of the complaint field in the historical complaint forms of the historical insurance cases and the number of historical complaint forms, and determine the target complaint field from the complaint fields based on the field weight.

[0054] It should be noted that the complaint form includes various fields for recording the complaint content, and different complaint fields have different effects on complaint escalation. For example, in the case of complaint escalation, some high-frequency fields may also exist, such as "I want to complain about this company" or "I do not accept the current solution, I want to continue to complain", etc. Therefore, the weight of the complaint field can be used to characterize the influence of the field on the complaint escalation. For example, the larger the field weight, the greater the probability that the internal complaint form containing the field will be upgraded to a regulatory complaint form.

[0055] Furthermore, since there are too many complaint fields, the complaint fields may contain useless fields, such as the name of the complaining customer and the agent responsible for the complaint. Therefore, in order to reduce the amount of resources required for training, the target complaint field can be determined from the complaint field based on the field weight, and the complaint risk prediction model can be trained using the field weight of the target complaint field to avoid the impact of multiple useless interference fields on the prediction results, and at the same time, reduce the amount of data required for training.

[0056] Step 140: Train the complaint risk prediction model according to the field weight of the target complaint field.

[0057] Specifically, the complaint risk prediction model can be trained according to the field weight of the target complaint field to obtain the complaint risk prediction model when the training is successful.

[0058] The embodiment of the present invention provides a training method for a complaint risk prediction model, the method comprising: obtaining multiple historical complaint sheets of historical insurance cases, wherein the historical complaint sheets include complaint time and complaint fields, and the complaint fields include complaint level fields; marking the historical insurance cases as positive samples or negative samples according to the complaint time and complaint level fields of the historical complaint sheets of the historical insurance cases, wherein there is a progressive relationship between the complaint levels between the historical complaint sheets corresponding to the positive samples, and the complaint levels represent the severity of the complaints; for the historical insurance cases corresponding to the positive samples and the negative samples, determining the field weight of the complaint field according to the field value of the complaint field of the historical complaint sheets of the historical insurance cases and the number of historical complaint sheets, determining the target complaint field from the complaint field according to the field weight; training the complaint risk prediction model according to the field weight of the target complaint field. Specifically, by accurately marking the positive samples in the historical insurance cases and determining the target complaint field from the complaint field according to the field weight, the accuracy of the model training can be improved, and the resource consumption of the training process can be reduced. Furthermore, the complaint risk prediction model trained by the method can accurately predict the complaint risk and improve the processing efficiency of the complaint sheets.

[0059] Embodiment 2

[0060] Figure 2 A flowchart of a method for training a complaint risk prediction model is provided in the second embodiment of the present invention. This embodiment is based on the technical solutions of the above embodiments and further limits the method for determining the field weights of the complaint fields.

[0061] like Figure 2 As shown, including:

[0062] Step 210: Obtain multiple historical complaint forms of historical insurance cases, wherein the historical complaint forms include complaint time and complaint fields, and the complaint fields include complaint level fields.

[0063] Step 220: Mark the historical insurance case as a positive sample or a negative sample based on the complaint time and complaint level fields of the historical complaint form of the historical insurance case, wherein there is a progressive relationship of complaint levels between the historical complaint forms corresponding to the positive samples, and the complaint level represents the severity of the complaint.

[0064] Step 230: For the historical insurance cases corresponding to the positive samples and negative samples, the historical complaint forms are binned according to the field value of the complaint field to generate data boxes.

[0065] Step 240: Count the number of historical complaint forms corresponding to the positive samples and the number of historical complaint forms corresponding to the negative samples in each of the data boxes.

[0066] Step 250: Determine the field weight of the complaint field according to the number of historical complaint forms corresponding to the positive samples and the number of historical complaint forms corresponding to the negative samples in each of the data boxes.

[0067] Steps 230-250 are the process of determining the field weight of the complaint field.

[0068] Specifically, historical complaint forms can be binned according to the field value of the complaint field to generate data boxes, so that the field value of the complaint field in each data box is different, and then the field weight of the complaint field is determined by calculating the number of samples corresponding to the complaint field in each data box. It can be understood that if the number of positive samples in field A is large and the number of negative samples is small, it indicates that the complaint field of this field value is the key field for complaint escalation, and the field weight of this complaint field is large.

[0069] Optionally, the field weight can be determined by the following formula:

[0070]

[0071] Among them, WOE is the field weight, and are the number of positive samples and the total number of positive samples in the data box respectively; and is the number of negative samples and the total number of negative samples of the data item.

[0072] Step 260: Determine a target complaint field from the complaint fields according to the field weight.

[0073] Optionally, if the field weight of the complaint field is greater than a preset weight threshold, the complaint field is determined as a target complaint field.

[0074] Specifically, the complaint fields can be screened by preset weight thresholds to determine the target complaint fields that have a greater impact on complaint escalation, thereby improving the accuracy of complaint prediction and reducing the amount of training data.

[0075] Optionally, the target complaint fields may be screened by one or more methods of correlation analysis and collinearity analysis, which are not limited here.

[0076] Step 270: Train the complaint risk prediction model according to the field weight of the target complaint field.

[0077] Optionally, step 270 includes:

[0078] The field weight of the target complaint field of the historical complaint form is input into the complaint risk prediction model to be trained.

[0079] The complaint risk prediction model to be trained is used to determine the complaint upgrade probability of the corresponding historical complaint form according to the field weights of each target complaint field of the historical complaint form.

[0080] Optionally, the complaint risk prediction model may include an xgboost (eXtreme Gradient Boosting) algorithm or an LR (Logistic Regression) algorithm, and then the probability of complaint upgrade of the corresponding historical complaint form may be determined according to the field weights of each target complaint field of the historical complaint form. The specific implementation and usage of the algorithm are not limited here.

[0081] Optionally, you can Determine the complaint upgrade probability of historical complaint tickets, where lrp is the complaint upgrade probability, is the intercept, is the logistic regression coefficient, and x is the field weight of the complaint field.

[0082] The predicted complaint probability of historical insurance cases is determined based on the complaint escalation probability of each historical complaint form.

[0083] Optionally, the corresponding historical insurance case is determined based on the complaint ticket number of the historical complaint ticket; and the predicted complaint probability of the historical insurance case is determined based on the complaint ticket upgrade probability of each historical complaint ticket of the historical insurance case.

[0084] Specifically, the average complaint escalation probability of historical complaint sheets of historical insurance cases may be determined as the predicted complaint probability of historical insurance cases; or the maximum complaint escalation probability may be determined as the predicted complaint probability of historical insurance cases.

[0085] The true complaint probability is determined according to the annotated type of the historical insurance cases. If the error value between the predicted complaint probability and the true complaint probability is less than a preset error threshold, the training ends.

[0086] Optionally, during the training process, an error function is preset, and an error value can be determined based on the error function, the true complaint probability, and the predicted complaint probability. Further, if the error value is less than a preset error threshold, the training ends. The annotation types of historical insurance cases include positive samples or negative samples, such as the true complaint probability of a positive sample is 1, and the true complaint probability of a negative sample is 0.

[0087] The embodiment of the present invention provides a method for training a complaint risk prediction model. Through the method of this embodiment, the field weight of the target complaint field can be accurately determined, and then the complaint risk prediction model can be trained according to the field weight of the target complaint field. Complaint prediction can be performed on the current complaint form based on the successfully trained complaint risk prediction model, and the complaint escalation probability of the current complaint form can be accurately determined, which can improve the accuracy of complaint risk prediction in the insurance field.

[0088] Embodiment 3

[0089] Figure 3 This is a flowchart of a complaint risk determination method provided in Embodiment 3 of the present invention. The method is applied to the complaint risk prediction model described in the embodiment of the present invention and predicts the complaint risk of the current insurance case, specifically including:

[0090] Step 310: Obtain the current complaint form and field weight table of the current insurance case, wherein the current complaint form includes complaint fields, and the field weight table is determined based on the complaint fields of historical complaint forms and the number of historical complaint forms.

[0091] Among them, the current complaint form is a complaint form generated for the current insurance case based on the customer's complaint needs.

[0092] Among them, the field weight table includes the field weights of each complaint field. In the training method of the complaint risk prediction model described in the embodiment of the present invention, a method for determining the field weights of the complaint fields is included. After determining the field weights of each complaint field, the field weights of each complaint field can be recorded and a field weight table can be generated. Furthermore, during the use stage of the complaint risk prediction model, the target complaint field and the corresponding field weight of the current complaint form can be directly determined based on the field weight table.

[0093] Step 320: Determine the target complaint field and the corresponding field weight according to the complaint field and field weight table of the current complaint form.

[0094] Specifically, the field weight of each complaint field in the current complaint form can be determined by querying the field weight table. Furthermore, when the field weight of a complaint field is greater than a preset threshold, it can be determined as a target complaint field.

[0095] Step 330: Based on the complaint risk prediction model, determine the complaint escalation probability of the current insurance case according to the field weight of the target complaint field.

[0096] Specifically, after the field weight of the target complaint field is input into the complaint risk prediction model, the complaint risk prediction model will output the complaint escalation probability of the current complaint case.

[0097] Among them, the complaint escalation probability is the probability of an internal complaint being upgraded to a regulatory complaint. The greater the complaint escalation probability, the more urgent it is to handle the insurance case as soon as possible to avoid complaint escalation and cause a wider range of complaint impacts.

[0098] Step 340: Determine the complaint risk of the current insurance case based on the complaint escalation probability.

[0099] Specifically, different complaint escalation probabilities can be preset to correspond to different complaint risk levels, thereby helping users determine the complaint risk of the current complaint form. For example, complaint forms with high complaint risks can be handled by experienced personnel to improve the handling efficiency of complaint forms.

[0100] The complaint risk determination method provided in the embodiment of the present invention can determine the target complaint field and the corresponding field weight of the current complaint form based on the complaint field and the field weight table in the current complaint form, and then based on the complaint risk prediction model, determine the complaint escalation probability of the current insurance case according to the field weight of the target complaint field, and determine the complaint risk of the current insurance case through the complaint escalation probability. Through the method of the embodiment of the present invention, the accuracy of complaint risk prediction can be improved, and the processing efficiency of complaint forms can be improved.

[0101] Embodiment 4

[0102] Figure 4 A structural schematic diagram of a complaint risk prediction model training device provided in an embodiment of the present invention, which can be used to execute any complaint risk prediction model training method described in the embodiments of the present invention.

[0103] like Figure 4 Including:

[0104] An acquisition module 410 is used to acquire a plurality of historical complaint sheets of historical insurance cases, wherein the historical complaint sheets include a complaint time and a complaint field, and the complaint field includes a complaint level field;

[0105] A labeling module 420 is used to label the historical insurance case as a positive sample or a negative sample according to the complaint time and complaint level field of the historical complaint sheet of the historical insurance case, wherein there is a progressive relationship of complaint levels between the historical complaint sheets corresponding to the positive samples, and the complaint level represents the severity of the complaint;

[0106] A target complaint field determination module 430 is used to determine the field weight of the complaint field for the historical insurance cases corresponding to the positive samples and the negative samples according to the field value of the complaint field of the historical complaint sheets of the historical insurance cases and the number of historical complaint sheets, and determine the target complaint field from the complaint fields according to the field weight;

[0107] The training module 440 is used to train the complaint risk prediction model according to the field weight of the target complaint field.

[0108] The device of the embodiment of the present invention can improve the accuracy of model training and reduce resource consumption in the training process by accurately marking positive samples in historical insurance cases and determining the target complaint field from the complaint field according to the field weight. Furthermore, by training a successfully complaint risk prediction model, the accuracy of complaint risk prediction in the insurance field can be improved.

[0109] Optionally, the labeling module 420 is specifically used to determine the historical insurance case as a positive sample if the complaint level fields of at least two historical complaint forms for the same historical insurance case are characterized by a progressive relationship in the complaint levels, and the complaint time interval between the two historical complaint forms is less than a preset interval, otherwise it is determined as a negative sample.

[0110] Optionally, the target complaint field determination module 430 includes:

[0111] Field weight determination unit and target complaint field determination unit

[0112] Optionally, the field weight determination unit includes: a box division unit, used to divide the historical complaint forms into boxes according to the field value of the complaint field to generate data boxes; a statistical unit, used to count the number of historical complaint forms corresponding to the positive samples and the number of historical complaint forms corresponding to the negative samples in each of the data boxes; a weight calculation unit, used to determine the field weight of the complaint field according to the number of historical complaint forms corresponding to the positive samples and the number of historical complaint forms corresponding to the negative samples in each of the data boxes.

[0113] The target complaint field determination unit is used to determine the complaint field as the target complaint field if the field weight of the complaint field is greater than a preset weight threshold.

[0114] Optionally, the training module 440 includes:

[0115] An input unit, used for inputting the field weight of the target complaint field of the historical complaint form into the complaint risk prediction model to be trained;

[0116] The probability determination unit is used to determine the predicted complaint probability of historical insurance cases based on the complaint escalation probability of each historical complaint form.

[0117] A judgment unit is used to determine the true complaint probability according to the marked type of the historical insurance case. If the error value between the predicted complaint probability and the true complaint probability is less than a preset error threshold, the training ends.

[0118] Optionally, the probability determination unit is specifically used to determine the corresponding historical insurance case according to the complaint ticket number of the historical complaint ticket;

[0119] The predicted complaint probability of the historical insurance case is determined based on the complaint escalation probability of each historical complaint form of the historical insurance case.

[0120] The training device for the complaint risk prediction model provided in the embodiment of the present invention can execute the training method for the complaint risk prediction model provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0121] Embodiment 5

[0122] Figure 5 The present invention also provides a schematic diagram of the structure of a complaint risk determination device, which is used to execute any complaint risk determination method described in the embodiments of the present invention. Specifically, Figure 5 As shown, including:

[0123] The case acquisition module 510 is used to acquire the current complaint form and field weight table of the current insurance case, wherein the current complaint form includes complaint fields, and the field weight table is determined according to the complaint fields of historical complaint forms and the number of historical complaint forms;

[0124] The weight determination module 520 is used to determine the target complaint field and the corresponding field weight according to the complaint field and field weight table of the current complaint form;

[0125] An input module 530 is used to determine the complaint escalation probability of the current insurance case according to the field weight of the target complaint field based on the complaint risk prediction model;

[0126] The risk determination module 540 is used to determine the complaint risk of the current insurance case according to the complaint escalation probability.

[0127] The complaint risk determination device provided in the embodiment of the present invention can determine the target complaint field and the corresponding field weight of the current complaint form based on the complaint field and the field weight table in the current complaint form, and then based on the complaint risk prediction model, determine the complaint escalation probability of the current insurance case according to the field weight of the target complaint field, and determine the complaint risk of the current insurance case through the complaint escalation probability. Through the method of the embodiment of the present invention, the accuracy of complaint risk prediction can be improved and the processing efficiency of complaint forms can be improved.

[0128] Embodiment 6

[0129] Figure 6A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, 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 processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0130] like Figure 6 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. The input / output (I / O) interface 15 is also connected to the bus 14.

[0131] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0132] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the training method of the complaint risk prediction model or the complaint risk determination method.

[0133] In some embodiments, the training method of the complaint risk prediction model or the complaint risk determination method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the training method of the complaint risk prediction model or the complaint risk determination method described above may be executed. Alternatively, in other embodiments, the processor 11 may be configured to execute the training method of the complaint risk prediction model or the complaint risk determination method in any other appropriate manner (for example, by means of firmware).

[0134] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0135] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0136] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0137] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0138] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0139] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0140] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0141] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for training a complaint risk prediction model, characterized in that: include: Acquire multiple historical complaint sheets of historical insurance cases, wherein the historical complaint sheets include complaint time and complaint fields, and the complaint fields include complaint level fields; According to the complaint time and complaint level fields of the historical complaint sheets of the historical insurance cases, the historical insurance cases are marked as positive samples or negative samples, wherein there is a progressive relationship of complaint levels between the historical complaint sheets corresponding to the positive samples, and the complaint levels represent the severity of the complaints; For the historical insurance cases corresponding to the positive samples and the negative samples, determine the field weight of the complaint field according to the field value of the complaint field of the historical complaint sheet of the historical insurance case and the number of historical complaint sheets, and determine the target complaint field from the complaint field according to the field weight; The complaint risk prediction model is trained according to the field weight of the target complaint field.

2. The method according to claim 1, characterized in that The marking of the historical insurance case as a positive sample or a negative sample according to the complaint time and complaint level fields of the historical complaint sheet of the historical insurance case includes: If there are at least two historical complaint forms for the same historical insurance case, and the complaint level fields of which indicate a progressive relationship in complaint level, and the complaint time interval between the two historical complaint forms is less than a preset interval, then the historical insurance case is determined as a positive sample, otherwise it is determined as a negative sample.

3. The method according to claim 1, characterized in that Determining the field weight of the complaint field according to the field value of the complaint field of the historical complaint sheet of the historical insurance case and the number of historical complaint sheets includes: According to the field value of the complaint field, the historical complaint form is divided into boxes to generate data boxes; Counting the number of historical complaint forms corresponding to the positive samples and the number of historical complaint forms corresponding to the negative samples in each of the data boxes; The field weight of the complaint field is determined according to the number of historical complaint sheets corresponding to the positive samples and the number of historical complaint sheets corresponding to the negative samples in each of the data boxes.

4. The method according to claim 3, characterized in that The determining a target complaint field from the complaint fields according to the field weights includes: If the field weight of the complaint field is greater than a preset weight threshold, the complaint field is determined as a target complaint field.

5. The method according to claim 1, characterized in that The step of training the complaint risk prediction model according to the field weight of the target complaint field includes: Inputting the field weight of the target complaint field of the historical complaint form into the complaint risk prediction model to be trained; By using the complaint risk prediction model to be trained, the probability of complaint upgrade of the corresponding historical complaint form is determined according to the field weights of each target complaint field of the historical complaint form; Determine the predicted complaint probability of historical insurance cases according to the complaint escalation probability of each historical complaint sheet; The actual complaint probability is determined according to the annotated type of the historical insurance cases. If the error value between the predicted complaint probability and the actual complaint probability is less than a preset error threshold, the training ends.

6. The method according to claim 5, characterized in that Determining the predicted complaint probability of historical insurance cases according to the complaint ticket upgrade probability of each historical complaint ticket includes: Determine the corresponding historical insurance case according to the complaint number of the historical complaint form; The predicted complaint probability of the historical insurance case is determined based on the complaint escalation probability of each historical complaint form of the historical insurance case.

7. A method for determining complaint risk, characterized in that: include: Obtaining a current complaint form and a field weight table of a current insurance case, wherein the current complaint form includes complaint fields, and the field weight table is determined according to complaint fields of historical complaint forms and the number of historical complaint forms; Determine the target complaint field and the corresponding field weight according to the complaint field and field weight table of the current complaint form; Based on the complaint risk prediction model, the complaint escalation probability of the current insurance case is determined according to the field weight of the target complaint field; The complaint risk of the current insurance case is determined based on the complaint escalation probability.

8. A training device for a complaint risk prediction model, characterized in that: include: An acquisition module, used for acquiring a plurality of historical complaint sheets of historical insurance cases, wherein the historical complaint sheets include a complaint time and a complaint field, and the complaint field includes a complaint level field; A labeling module, used for labeling the historical insurance cases as positive samples or negative samples according to the complaint time and complaint level fields of the historical complaint sheets of the historical insurance cases, wherein there is a progressive relationship of complaint levels between the historical complaint sheets corresponding to the positive samples, and the complaint levels represent the severity of the complaints; A target complaint field determination module is used to determine the field weight of the complaint field for the historical insurance cases corresponding to the positive samples and the negative samples according to the field value of the complaint field of the historical complaint sheets of the historical insurance cases and the number of historical complaint sheets, and determine the target complaint field from the complaint fields according to the field weight; A training module is used to train the complaint risk prediction model according to the field weight of the target complaint field.

9. A complaint risk determination device, characterized in that: include: A case acquisition module, used to acquire a current complaint form and a field weight table of a current insurance case, wherein the current complaint form includes a complaint field, and the field weight table is determined according to the complaint fields of historical complaint forms and the number of historical complaint forms; A weight determination module, used to determine the target complaint field and the corresponding field weight according to the complaint field and field weight table of the current complaint form; An input module, used to determine the complaint escalation probability of the current insurance case according to the field weight of the target complaint field based on the complaint risk prediction model; The risk determination module is used to determine the complaint risk of the current insurance case based on the complaint escalation probability.

10. An electronic device, characterized in that: The electronic device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the training method of the complaint risk prediction model described in any one of claims 1-6 or the complaint risk determination method described in claim 7.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, which are used to enable a processor to implement the training method for the complaint risk prediction model described in any one of claims 1 to 6 or the complaint risk determination method described in claim 7 when executed.

Images