User complaint early warning method, device, equipment and storage medium

Abstract

The application discloses a user complaint early warning method and device, equipment and storage medium, the method comprises the following steps: obtaining user complaint information; inputting the user complaint information into a preset naive Bayes classification model to obtain a complaint classification result; determining a complained business according to the complaint classification result and performing backtracking matching on the complained business through a preset business subscription big data model to obtain historical business handling information of the complained business; generating a business early warning work order according to the historical business handling information and performing user complaint early warning based on the business early warning work order. Since the user complaint information is classified through the naive Bayes classification model, compared with artificial classification, the workload of repeated inquiries of customer service is reduced. Meanwhile, the historical business handling information is obtained by backtracking matching on the complained business through the business subscription big data model, then the information is analyzed to generate a business early warning work order, and the complaint early warning is performed, so that early warning of user complaints can be realized, and the risk of similar complaints is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of customer service, and in particular to a user complaint early warning method, device, equipment and storage medium. BACKGROUND

[0002] In the field of mobile communication, the existing complaint user classification and management method is: relying on customer service personnel to repeatedly confirm the complaint content with the complaint user and manually classifying the complaint business, content, etc. according to the customer description content, and then according to the manually classified complaint content and the complaint business, organization, etc. The corresponding person in charge feeds back and handles the complaint, and finally solves the user's complaint demand.

[0003] The disadvantages of the prior art are mainly as follows:

[0004] 1. The classification process of user complaints is too lengthy and inefficient, and the large number of daily complaint work orders makes it impossible to perform full-quantity complaint work order customer classification;

[0005] 2. For customer complaints, only one complaint can be found and managed, and it is impossible to prevent user complaints from occurring, and the effect of reducing the ratio of complaints to services is not good;

[0006] 3. The complaint management after the fact is to maintain and manage after the user has actually been dissatisfied, and the effect of helping to improve the overall user satisfaction is not good.

[0007] From the above, the existing user complaint processing method has the technical problems of consuming a large amount of manpower and material resources, and being unable to effectively warn user complaints, and the overall user satisfaction is low.

[0008] The above content is only used to assist in understanding the technical solutions of the present application, and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0009] The main purpose of the present application is to provide a user complaint early warning method, device, equipment and storage medium, which aims to solve the technical problems of the existing user complaint processing method consuming a large amount of manpower and material resources, and being unable to effectively warn user complaints, and the overall user satisfaction being low.

[0010] To achieve the above purpose, the present application provides a user complaint early warning method, which comprises the following steps:

[0011] Obtaining user complaint information in a preset period;

[0012] Inputting the user complaint information into a preset Naive Bayes classification model to obtain a complaint classification result;

[0013] determine a complained business according to the complaint classification result, and perform backtracking matching on the complained business through a preset business subscription big data model to obtain historical business handling information of the complained business;

[0014] generate a business early warning work order according to the historical business handling information, and perform user complaint early warning based on the business early warning work order.

[0015] Preferably, before the step of obtaining user complaint information in a preset time period, the method further comprises:

[0016] obtaining user historical complaint information, and extracting keywords of a preset dimension from the user historical complaint information;

[0017] dividing the keywords into feature attributes, and constructing model training samples according to the division result;

[0018] inputting the model training samples into an initial naive Bayes classification model for training to obtain a naive Bayes classification model to be tuned;

[0019] performing parameter tuning on the naive Bayes classification model to be tuned according to a model training result to obtain a preset naive Bayes classification model.

[0020] Preferably, the step of determining a complained business according to the complaint classification result, and performing backtracking matching on the complained business through a preset business subscription big data model to obtain historical business handling information of the complained business comprises:

[0021] determining a complained business according to the complaint classification result;

[0022] obtaining a complaint user identifier corresponding to the user complaint information, and searching for a corresponding preset business subscription big data model in a preset model database according to the complaint user identifier;

[0023] performing backtracking matching on the complained business through the preset business subscription big data model to obtain historical business handling information of the complained business.

[0024] Preferably, the step of performing backtracking matching on the complained business through the preset business subscription big data model to obtain historical business handling information of the complained business comprises:

[0025] obtaining historical business handling records of a corresponding complaint user through the preset business subscription big data model;

[0026] obtaining a business level classification result corresponding to the complained business;

[0027] The historical service handling information is obtained by matching the service level classification result with the historical service handling record.

[0028] Preferably, the service early warning work order comprises a product service risk early warning work order.

[0029] The step of generating a service early warning work order according to the historical service handling information and performing user complaint early warning based on the service early warning work order comprises:

[0030] The service handling time information corresponding to the complained service in the historical service handling information is extracted;

[0031] The specific service content information is determined according to the service handling time information;

[0032] The product service risk early warning work order is generated according to the specific service content information, and user complaint early warning is performed based on the product service risk early warning work order.

[0033] Preferably, the step of generating a product service risk early warning work order according to the specific service content information and performing user complaint early warning based on the product service risk early warning work order comprises:

[0034] The service type to which the complained service belongs is determined according to the specific service content information;

[0035] The number of user complaints for a product service corresponding to the service type under different service handling organizations within a preset time period is obtained;

[0036] When the number of user complaints exceeds a first preset threshold, the process design information corresponding to the product service is obtained;

[0037] The product service risk early warning work order is generated according to the process design information, and user complaint early warning is performed based on the product service risk early warning work order.

[0038] Preferably, the service early warning work order further comprises a channel service handling abnormality early warning work order.

[0039] After the step of determining the specific service content information according to the service handling time information, the method further comprises:

[0040] The service handling organization corresponding to the complained service is determined according to the specific service content information;

[0041] All user complaint information within a preset time period is obtained, and the number of user complaints corresponding to the service handling organization is determined according to the user complaint information;

[0042] When the number of user complaints exceeds a second preset threshold, it is determined that the service handling organization has service handling abnormalities, and a channel service handling abnormality early warning work order is output;

[0043] Based on the channel service handling abnormality early warning work order, user complaint early warning is performed.

[0044] In addition, to achieve the above-mentioned purpose, the present application also provides a user complaint early warning device, which comprises:

[0045] An information acquisition module is configured to acquire user complaint information in a preset period of time;

[0046] An information classification module is configured to input the user complaint information into a preset Naive Bayes classification model to obtain a complaint classification result;

[0047] A service matching module is configured to determine a complained service according to the complaint classification result, and perform backtracking matching on the complained service through a preset service subscription big data model to obtain historical service handling information of the complained service;

[0048] A complaint early warning module is configured to generate a service early warning work order according to the historical service handling information, and perform user complaint early warning based on the service early warning work order.

[0049] In addition, to achieve the above-mentioned purpose, the present application also provides a user complaint early warning device, which comprises a memory, a processor, and a user complaint early warning program stored in the memory and executable on the processor, wherein the user complaint early warning program is configured to implement the steps of the user complaint early warning method as described above.

[0050] In addition, to achieve the above-mentioned purpose, the present application also provides a storage medium, wherein the storage medium stores a user complaint early warning program, and the user complaint early warning program implements the steps of the user complaint early warning method as described above when executed by a processor.

[0051] This invention acquires user complaint information within a preset time period; inputs this information into a preset Naive Bayes classification model to obtain complaint classification results; identifies the complained-about service based on the classification results, and performs back-matching on the complained-about service using a preset service ordering big data model to obtain historical service processing information; generates a service warning work order based on the historical service processing information, and issues user complaint warnings based on the service warning work order. Because this invention automatically classifies user complaint information using a trained Naive Bayes classification model, it reduces the workload of repeated inquiries by customer service representatives compared to existing manual classification methods, saving significant human and material resources. Simultaneously, by using a service ordering big data model to back-match the complained-about service identified by the complaint classification results to obtain historical service processing information, and then analyzing this information to generate a service warning work order for early warning, this invention enables early warning of user complaints and reduces the risk of similar complaints. Attached Figure Description

[0052] Figure 1 This is a schematic diagram of the structure of a user complaint early warning device in the hardware operating environment involved in the embodiments of the present invention;

[0053] Figure 2 This is a flowchart illustrating the first embodiment of the user complaint early warning method of the present invention;

[0054] Figure 3 This is a flowchart illustrating the second embodiment of the user complaint early warning method of the present invention;

[0055] Figure 4 This is a flowchart illustrating the third embodiment of the user complaint early warning method of the present invention;

[0056] Figure 5 This is a structural block diagram of the first embodiment of the user complaint early warning device of the present invention.

[0057] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0058] It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention.

[0059] Reference Figure 1 , Figure 1 This is a schematic diagram of the user complaint early warning device structure in the hardware operating environment involved in the embodiments of the present invention.

[0060] like Figure 1As shown, the user complaint early warning device can include a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to realize the connection and communication between the components. The user interface 1003 can include a display, an input unit such as a keyboard, and can also include a standard wired interface, a wireless interface. The network interface 1004 can optionally include a standard wired interface, a wireless interface (such as a wireless fidelity (WIreless-FIdelity, WI-FI) interface). The memory 1005 can be a high-speed random access memory (RAM) memory, or a stable non-volatile memory (Non-Volatile Memory, NVM) such as a disk memory. The memory 1005 can also be a storage device independent of the aforementioned processor 1001.

[0061] Those skilled in the art can understand that Figure 1 The structure shown in the figure does not constitute a limitation on the user complaint early warning device, and can include more or fewer components than the figure, or combine certain components, or different component arrangements.

[0062] As Figure 1 As shown, the memory 1005 as a storage medium can include an operating system, a data storage module, a network communication module, a user interface module, and a user complaint early warning program.

[0063] In Figure 1 As shown in the user complaint early warning device, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the user complaint early warning device of the application can be arranged in the user complaint early warning device, and the user complaint early warning device calls the user complaint early warning program stored in the memory 1005 through the processor 1001, and executes the user complaint early warning method provided by the embodiment of the application.

[0064] The embodiment of the application provides a user complaint early warning method, which refers to Figure 2 , Figure 2 The flowchart of the first embodiment of the user complaint early warning method of the application.

[0065] In this embodiment, the user complaint early warning method includes the following steps:

[0066] Step S10: Obtain user complaint information in a preset period.

[0067] It should be noted that the subject of the method of the embodiment can be a computing service device with data processing, network communication and program running functions, such as a smart phone, a tablet computer and a personal computer, etc.; it can also be other electronic devices with the same or similar functions or the above-mentioned user complaint early warning device. The embodiment and the following embodiments will be described taking the user complaint early warning device (hereinafter referred to as the early warning device) as an example.

[0068] It can be understood that the above-mentioned preset time period can be a period of time set according to actual conditions, such as one month, three months, half a year, one year, etc. The user complaint information can be business complaint information fed back to the communication service provider by the user through different channels (such as manual service telephone, short message, outbound system, etc.). The complaint information can contain information such as the complained business, the complaint type and the business acceptance type. Among them, the business acceptance type can be divided into three levels, such as first-level classification (such as home broadband complaint, business complaint, wireless network complaint, etc.), second-level classification (such as traffic, contract, voice, business process, service problem, etc.), and third-level classification (such as traffic package, no additional charge for additional volume, package migration, voice contract, physical contract, terminal contract, rights and interests business, voice business, etc.).

[0069] Step S20: inputting the user complaint information into a preset Naive Bayes classification model to obtain a complaint classification result;

[0070] It should be noted that the Naive Bayes model is a simple and fast classification algorithm, which is usually suitable for very high-dimensional data sets. Because of its fast running speed and few adjustable parameters, it is very suitable for providing a fast basic solution for classification problems.

[0071] In this step, the preset Naive Bayes classification model can be a Naive Bayes classification model obtained by pre-training the model according to the training sample and optimizing the parameters. This model can be used to classify the user complaint information according to the set dimensions (such as business type, complaint type, business acceptance, etc.) to obtain the complaint classification result.

[0072] It should be understood that for the trained preset Naive Bayes classification model, the input is the user complaint information, and the output is the classification result containing different levels (i.e. the above-mentioned complaint classification result), for example, user A, first-level classification: business complaint, second-level classification: traffic, third-level classification: traffic volume increase without additional charge, outbound complaint: yes, repeated complaint: no, and unaware customization: no.

[0073] Further, in order to ensure the accuracy of the model classification, the user complaint early warning method of the embodiment further comprises a model training process before the above-mentioned step S10, which specifically includes:

[0074] Step S01: obtaining user historical complaint information, and extracting keywords of preset dimensions from the user historical complaint information;

[0075] It should be noted that the user historical complaint information can be user complaints of different channels in a certain region, area or even nationwide received by a communication service provider in the past period of time (the specific duration is not limited). The preset dimensions include but are not limited to three dimensions of service type, complaint type and service handling type. Correspondingly, the keywords extracted from each user complaint according to these dimensions will include service type keywords, complaint type keywords, service handling type keywords, etc.

[0076] Step S02: dividing the keywords into feature attributes, and constructing model training samples according to the division results;

[0077] It should be understood that the feature attribute can be the category to which each keyword belongs (which can be artificially defined). In a specific implementation, the feature attributes of each keyword can be appropriately divided first, and then a part of the divided keywords are manually classified to form model training samples.

[0078] Step S03: inputting the model training samples into an initial naive Bayes classification model for training to obtain a naive Bayes classification model to be tuned;

[0079] It should be understood that the model needs to undergo many times of iterative training from the initial training to the end of training (i.e. model convergence), and the purpose of iterative training is also to optimize the parameters in the model of each training, so that the finally trained model is optimal.

[0080] Step S04: parameter tuning of the naive Bayes classification model to be tuned according to the model training results to obtain a preset naive Bayes classification model.

[0081] It should be noted that the model training results can be the frequency of each category in the training samples calculated by each generation of the trained naive Bayes classification model and the conditional probability of each feature attribute division to each category. According to these training results, the parameter tuning of each generation of the naive Bayes classification model can be performed until the model converges, and the preset naive Bayes classification model is obtained.

[0082] In a specific implementation, in order to realize effective early warning of user complaints, the early warning device can input the user complaint information into the preset naive Bayes classification model to obtain a complaint classification result.

[0083] Step S30: determining the complained service according to the complaint classification result, and performing backtracking matching on the complained service through a preset service subscription big data model to obtain historical service handling information of the complained service;

[0084] It should be noted that the preset service subscription big data model can be a service subscription model constructed for each user, and the historical service handling records of the user are recorded in the model, for example, the user A opened a traffic refueling package in January 1, 2020, and upgraded the traffic package in June 1, 2020.

[0085] In a specific implementation, the complained service can be determined according to the complaint classification result, and then the history service is backtracked (the backtracking time can be set by oneself) through the preset service subscription big data model according to the complained service. For example, according to the complaint classification result corresponding to the user A, it is determined that the complained service is traffic service, and then the traffic service subscription records of the user A in the past two years are backtracked through the service subscription big data model corresponding to the user A, so as to determine the historical service handling information of the complained service, which includes service handling time, service handling organization (or channel) and the like.

[0086] Step S40: generating a service warning work order according to the historical service handling information, and performing user complaint warning based on the service warning work order.

[0087] It should be understood that the user complaint generally comes from the dissatisfaction or unreasonable of the service that has been handled, so after the historical service handling information of a large number of complaint users is obtained, the warning device of the embodiment can analyze whether the user complaints are a minority or a majority according to the services complained by these users, and then different warning work orders are produced according to the analysis.

[0088] In actual situation, the warning work order can be divided into at least two categories, one is a product business risk warning work order in product dimension, and the other is a channel business handling abnormality warning work order in organization dimension. Among them, the product business risk warning work order is triggered by the unreasonable design process of the product business, and the channel business handling abnormality warning work order is triggered by the abnormality of the service handling organization.

[0089] In a specific implementation, after the pre-warning device determines the historical service handling information of all complained services according to the user complaint information in a preset period (for example, one month), the pre-warning device can determine the number of complaints of the same type of service under each service handling organization in the complained services according to the historical service handling information. If it is found that the number of complaints of each service handling organization exceeds a preset threshold, it is determined that the complained service may have an unreasonable design process. At this time, a product service risk pre-warning work order needs to be generated, and the pre-warning work order is first pushed to the person in charge corresponding to the service. After the person in charge receives the pre-warning work order, the process of the product service is checked and verified. If it is found that there is an error in the process, the process is corrected and repaired in a timely manner.

[0090] Of course, if the pre-warning device determines that the service handling organizations corresponding to the complained services are all handled by the same handling organization according to the historical service handling information, and the complained services have no special relationship with each other, it indicates that the service handling of the handling organization is abnormal. At this time, a channel service handling abnormality pre-warning work order is output. For the channel service handling abnormality pre-warning work order, the pre-warning device first pushes the channel service handling abnormality pre-warning work order to the channel manager, and the channel manager checks whether there is a service violation in the channel, locates the problem cause, and performs feedback processing, verification and rectification.

[0091] Further, in order to reduce the possibility of customer complaints from the root cause and improve customer satisfaction, the pre-warning device of the embodiment timely performs customer sentiment care and maintenance for customers in the same type of service or channel handling exception without complaints. For example, the product service with process error is optimized for business configuration to eliminate the product marketing that is not beneficial to the customer; the channel abnormal handling service is canceled for the customer, and the service handling fee is refunded.

[0092] The embodiment obtains user complaint information in a preset period of time, inputs the user complaint information into a preset Naive Bayes classification model to obtain a complaint classification result, determines a complained service according to the complaint classification result, and performs backtracking matching on the complained service through a preset service subscription big data model to obtain historical service handling information of the complained service, generates a service warning work order according to the historical service handling information, and performs user complaint warning based on the service warning work order. Since the embodiment classifies the user complaint information through the trained Naive Bayes classification model, compared with the existing manual classification mode, the workload of repeated inquiries of manual customer service is reduced, and a large amount of manpower and material resources is saved. Meanwhile, the embodiment performs backtracking matching on the complained service determined according to the complaint classification result through the service subscription big data model, obtains the historical service handling information, and then generates the service warning work order for warning after analyzing the information, so that early warning of user complaints can be realized, and the risk of similar complaints can be reduced.

[0093] Reference Figure 3 , Figure 3 FIG. 2 is a flowchart of a second embodiment of the user complaint warning method of the present application.

[0094] Based on the above first embodiment, in the present embodiment, the step S30 comprises:

[0095] Step S301: determining a complained service according to the complaint classification result;

[0096] It should be understood that the complaint classification result contains three levels of classification results, and judgment results such as whether the user has an upgrade complaint, a repeated complaint, an unknowingly customized high-risk complaint, etc. For example, user A, first-level classification: service complaint, second-level classification: traffic, third-level classification: traffic amount increase without price increase, outbound call complaint: yes, repeated complaint: no, unknowingly customized: no.

[0097] In a specific implementation, the warning device can determine the complained service according to different levels of classification results contained in the complaint classification result.

[0098] Step S302: obtaining a complaint user identifier corresponding to the user complaint information, and searching for a corresponding preset service subscription big data model in a preset model database according to the complaint user identifier;

[0099] It should be noted that the complaint user identifier can be an identifier information such as a registered mobile phone number, name, and ID card number of the user, which can distinguish different users. In order to facilitate management of the service subscription big data models of different users and improve query efficiency, a dynamic mapping of the user identifier of a user and the service subscription big data model of the user can be established in advance, so that the warning device can quickly search for the preset service subscription big data model according to the dynamic mapping.

[0100] Step S303: The complained service is backtracked and matched through the preset service subscription big data model to obtain the historical service handling information of the complained service.

[0101] It should be understood that the service subscription big data model records the historical service handling records of the user. When the user complaint occurs, the complained service can be tracked through the preset service subscription big data model according to the complained service of the user complaint, the historical service handling information of the complained service is obtained, and then the historical service handling information is used to determine which time node the user complaint is directed to the complained service, so as to analyze the real complaint cause.

[0102] In order to accurately determine the cause of the complaint, as an implementation manner, the early warning device in the embodiment can obtain the historical service handling record of the corresponding complaint user through the preset service subscription big data model; then obtain the service level classification result corresponding to the complained service; and then match the service level classification result according to the historical service handling record to obtain the historical service handling information of the complained service.

[0103] For example, the historical service handling record of the corresponding complaint user A in the past year is obtained according to the service subscription big data model of the user A as: “2020 January 1, …, open 10G traffic refueling package …, handled by artificial customer service”, “2020 June 1, …, handle the traffic plus package without price increase …, handled by B business hall” and “2020 December 1, …, cancel voice reminder …, handled by C business hall”. At this time, the early warning device obtains the service level classification result corresponding to the complained service as “user A, first-level classification: service complaint, second-level classification: traffic, third-level classification: traffic plus without price increase”, determines that the specific service complained by the user this time is “traffic plus without price increase” according to the third-level classification result in the service level classification result, and combines the above historical service handling record to determine that the historical service handling information of the complained service is “2020 June 1, …, handle the traffic plus package without price increase …, handled by B business hall”.

[0104] In a specific implementation, after the early warning device obtains the historical service handling information of each complained service, the cause of the complaint can be analyzed according to the historical service handling information of the complained service in the preset time period, and then the corresponding early warning work order is generated according to the cause.

[0105] The embodiment can accurately obtain the historical service handling information of the service complained by each complaint user in the above manner, and then analyze the causes of various complaints in a certain period of time based on a large amount of the historical service handling information, and generate early warning work orders in different situations according to the causes, thereby ensuring the reliability of the early warning work order generation and providing effective guarantee for subsequent complaint early warning.

[0106] Reference Figure 4 , Figure 4 The flowchart of the third embodiment of the user complaint early warning method of the application is shown.

[0107] Based on the first embodiment, the service early warning work order can include a product service risk early warning work order and a channel service handling abnormality early warning work order.

[0108] Therefore, in the embodiment, the step S40 can include:

[0109] Step S401: Extract the service handling time information corresponding to the complained service in the historical service handling information.

[0110] Step S402: Determine the specific service content information according to the service handling time information.

[0111] It should be understood that in actual cases, there is a case that the number of times of handling the complained service in the historical service handling information of the complaint user is large, for example, the number of times of handling the traffic plus volume without price increase of user A in the past year is three: "2020-06-01, …… handle the traffic plus volume without price increase package …… B business hall handling", "2020-07-01, …… handle the traffic plus volume without price increase package …… C business hall handling" and "2020-08-01, …… handle the traffic plus volume without price increase package …… F business hall handling". At this time, it is necessary to accurately determine whether the user is dissatisfied with which service handling to generate a complaint, so the early warning device needs to extract the service handling time information corresponding to the complained service in the historical service handling information, and then determine the specific service content information according to the service handling time information combined with the user complaint information (the user generally explains when, where and what service is handled in the complaint process), that is, the specific information of the service complained this time, such as service type, handling time, handling place, handling organization (or channel) and the like.

[0112] Step S403: Generate a product service risk early warning work order according to the specific service content information, and perform user complaint early warning based on the product service risk early warning work order.

[0113] In a specific implementation, after obtaining the specific service content information of all complained services, the early warning device can perform big data analysis on the information to obtain how many types of complained services there are in a preset time period, how many complained services of each type there are, and which service handling organization each complained service belongs to, and then generate a product service risk warning work order according to the big data analysis result and a predetermined early warning rule, and perform user complaint early warning based on the product service risk warning work order.

[0114] To make the early warning more targeted and effective, as an implementation, the early warning device in the embodiment can further perform the following steps:

[0115] Step S403: determining a service type to which the complained service belongs according to the specific service content information;

[0116] For example, according to the specific service content information, it is determined that the service type to which the complained service "traffic refueling package" belongs is "traffic service".

[0117] Step S404: obtaining a user complaint quantity of a product service corresponding to the service type under different service handling organizations in a preset time period;

[0118] Step S405: obtaining flow design information corresponding to the product service when the user complaint quantity exceeds a first preset threshold value;

[0119] The first preset threshold value can be set according to actual conditions, and the embodiment does not limit this.

[0120] Step S406: generating a product service risk warning work order according to the flow design information, and performing user complaint early warning based on the product service risk warning work order.

[0121] It can be understood that different service types correspond to different product services, for example, there are many traffic products corresponding to traffic service, such as unlimited traffic package, traffic refueling package, and limited traffic package. In order to ensure the coverage of early warning, the early warning device can collect and analyze the user complaint situation of various service products under the same service type, and then determine whether the service product with too many user complaints has an unreasonable flow design according to the analysis result, so as to correct and repair in time.

[0122] Further, for organization dimension early warning, the early warning device in the embodiment further performs the following steps:

[0123] Step S403': determining a service handling organization corresponding to the complained service according to the specific service content information;

[0124] It should be understood that the business handling organization can be the name or number of the business hall or agency handling the complained business for the user. In order to determine whether the complained business is due to the problem in the product design process or due to the abnormality of the business handling organization, the pre-warning device needs to acquire the information of the business handling organization corresponding to all the complained businesses in this embodiment.

[0125] Step S404': acquiring all the user complaint information in a preset time length, and determining the number of user complaints corresponding to the business handling organization according to the user complaint information;

[0126] After acquiring the information of the business handling organization corresponding to all the complained businesses, all the user complaint information in a preset time length (for example, one month or half a month) can be acquired, and then the number of user complaints corresponding to the business handling organization handling these user complaints is determined according to the user complaint information, and the total number of user complaints belonging to the unified business handling organization, that is, the number of user complaints, is determined.

[0127] Step S405': when the number of user complaints exceeds a second preset threshold, determining that the business handling organization has a business handling abnormality, and outputting a channel business handling abnormality pre-warning work order;

[0128] Step S406': performing user complaint pre-warning based on the channel business handling abnormality pre-warning work order.

[0129] In a specific implementation, when the pre-warning device detects that the number of user complaints exceeds the second preset threshold (the value is adjustable), it is determined that the business handling organization has a business handling abnormality, and a channel business handling abnormality pre-warning work order is outputted, and then user complaint pre-warning is performed based on the channel business handling abnormality pre-warning work order, for example, the channel business handling abnormality pre-warning work order is first pushed to the channel manager, the channel manager performs business handling abnormality verification, checks whether there is a business violation in the channel, locates the problem cause, and performs feedback processing, verification and rectification, cancels the handled business for the user who has handled the business in the business handling organization but has not complained, and informs and returns the business handling fee to prevent the occurrence of complaints.

[0130] The above-mentioned method can analyze the root cause of the complaint according to the historical business handling information corresponding to the complained business, generate a pre-warning work order, and then perform user complaint pre-warning according to the pre-warning work order, so that the complaint that has not occurred is found and managed in advance, and the customer satisfaction is improved.

[0131] In addition, the embodiment of the present application also provides a storage medium, and the storage medium stores a user complaint pre-warning program. When the user complaint pre-warning program is executed by a processor, the steps of the user complaint pre-warning method described above are implemented.

[0132] Referring to Figure 5 , Figure 5 is a structural block diagram of a first embodiment of a user complaint early warning device of the present application.

[0133] As Figure 5 shown, the user complaint early warning device proposed by the embodiment of the present application comprises:

[0134] An information acquisition module 501 is configured to acquire user complaint information in a preset time period;

[0135] An information classification module 502 is configured to input the user complaint information into a preset Naive Bayes classification model to obtain a complaint classification result;

[0136] A service matching module 503 is configured to determine a complained service according to the complaint classification result, and perform backtracking matching on the complained service through a preset service subscription big data model to obtain historical service handling information of the complained service;

[0137] A complaint early warning module 504 is configured to generate a service early warning work order according to the historical service handling information, and perform user complaint early warning based on the service early warning work order.

[0138] The embodiment acquires user complaint information in a preset time period, inputs the user complaint information into a preset Naive Bayes classification model to obtain a complaint classification result, determines a complained service according to the complaint classification result, and performs backtracking matching on the complained service through a preset service subscription big data model to obtain historical service handling information of the complained service, generates a service early warning work order according to the historical service handling information, and performs user complaint early warning based on the service early warning work order. Since the embodiment classifies user complaint information automatically through a trained Naive Bayes classification model, compared with the existing manual classification method, the workload of repeated inquiries of manual customer service is reduced, and a large amount of manpower and material resources is saved. Meanwhile, the embodiment performs backtracking matching on the complained service determined by the complaint classification result through a service subscription big data model to obtain historical service handling information, analyzes the information, generates a service early warning work order for early warning, can realize early warning of user complaints, and reduces the risk of similar complaints.

[0139] Based on the above-mentioned first embodiment of the user complaint early warning device of the present application, a second embodiment of the user complaint early warning device of the present application is proposed.

[0140] In the embodiment, the user complaint early warning device further comprises a model training module, configured to acquire user historical complaint information, and extract keywords of preset dimensions from the user historical complaint information; perform feature attribute division on the keywords, and construct model training samples according to the division result; input the model training samples into an initial Naive Bayes classification model for training, to obtain a Naive Bayes classification model to be tuned; and perform parameter tuning on the Naive Bayes classification model to be tuned according to a model training result, to obtain a preset Naive Bayes classification model.

[0141] Further, the service matching module 503 is further configured to determine a complained service according to the complaint classification result; acquire a complaint user identifier corresponding to user complaint information, and search for a corresponding preset service subscription big data model in a preset model database according to the complaint user identifier; and perform backtracking matching on the complained service through the preset service subscription big data model, to obtain historical service handling information of the complained service.

[0142] Further, the service matching module 503 is further configured to acquire historical service handling records of a corresponding complaint user through the preset service subscription big data model; acquire a service level classification result corresponding to the complained service; and match the service level classification result according to the historical service handling records, to obtain the historical service handling information of the complained service.

[0143] Further, the complaint early warning module 504 is further configured to extract service handling time information corresponding to the complained service in the historical service handling information; determine specific service content information according to the service handling time information; generate a product service risk early warning work order according to the specific service content information, and perform user complaint early warning based on the product service risk early warning work order.

[0144] Further, the complaint early warning module 504 is further configured to determine a service type to which the complained service belongs according to the specific service content information; acquire a number of user complaints for a product service corresponding to the service type under different service handling organizations within a preset time length; acquire process design information corresponding to the product service when the number of user complaints exceeds a first preset threshold; generate a product service risk early warning work order according to the process design information, and perform user complaint early warning based on the product service risk early warning work order.

[0145] Further, the complaint early warning module 504 is further configured to determine a service handling organization corresponding to the complained service according to the specific service content information; acquire all user complaint information within a preset time length, and determine a user complaint quantity corresponding to the service handling organization according to the user complaint information; when the user complaint quantity exceeds a second preset threshold, determine that the service handling organization has service handling abnormity, and output a channel service handling abnormity early warning work order; and perform user complaint early warning based on the channel service handling abnormity early warning work order.

[0146] Other embodiments or specific implementations of the user complaint early warning device can refer to the above-mentioned method embodiments, which will not be described here.

[0147] It should be noted that in this paper, the term "include", "contain" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such process, method, article or system. Without more limitations, the element defined by the sentence "including a" does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0148] The above-mentioned embodiment number of the application is only for description, not representing the advantages and disadvantages of the embodiments.

[0149] Through the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment method can be realized by software and necessary general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, an optical disk), and includes a plurality of instructions for making a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device) execute the method described in each embodiment of the present application.

[0150] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent flow transformation made by using the content of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A user complaint early warning method, characterized in that, The user complaint early warning method includes: Obtain user complaint information within a preset time period; The user complaint information is input into a preset Naive Bayes classification model to obtain the complaint classification result. The complaint-related business is determined based on the complaint classification results, and the complaint-related business is back-matched through a preset business ordering big data model to obtain the historical business processing information of the complaint-related business. The preset business ordering big data model is a business ordering model built for each user, and the preset business ordering big data model contains the user's historical business processing records. A business warning work order is generated based on the historical business processing information, and a user complaint warning is issued based on the business warning work order. Prior to the step of obtaining user complaint information within a preset time period, the method further includes: obtaining historical user complaint information and extracting keywords of a preset dimension from the historical user complaint information; classifying the keywords by feature attributes and constructing model training samples based on the classification results, wherein the feature attributes are the categories to which each keyword belongs; The training samples of the model are input into the initial Naive Bayes classification model for training to obtain multiple generations of Naive Bayes classification models to be tuned; and the frequency of occurrence of the category calculated by each generation of Naive Bayes classification model in the training samples and the conditional probability of each feature attribute for each category are determined, and the frequency of occurrence and the conditional probability are used as the model training results. Based on the training results of the model, the parameters of each generation of the Naive Bayes classification model are tuned until the model converges, thus obtaining the preset Naive Bayes classification model.

2. The user complaint early warning method as described in claim 1, characterized in that, The step of determining the complained-about service based on the complaint classification results and performing a retrospective matching of the complained-about service using a preset service ordering big data model to obtain the historical service processing information of the complained-about service includes: The business being complained about is determined based on the complaint classification results; Obtain the complaint user identifier corresponding to the user complaint information, and search for the corresponding preset business order big data model in the preset model database based on the complaint user identifier; By using a pre-set big data model for business ordering, the complained-about business is back-matched to obtain historical business processing information for the complained-about business.

3. The user complaint early warning method as described in claim 2, characterized in that, The step of retrospectively matching the complained-about business using a preset business order big data model to obtain the historical business processing information of the complained-about business includes: The historical business processing records of the corresponding complainant user are obtained through the preset business order big data model. Obtain the business level classification result corresponding to the complained business; The historical business processing records are matched with the business level classification results to obtain the historical business processing information of the complained business.

4. The user complaint early warning method as described in claim 3, characterized in that, The business early warning work order includes: product business risk early warning work order; The steps of generating a business early warning work order based on the historical business processing information and issuing a user complaint early warning based on the business early warning work order include: Extract the business processing time information corresponding to the complained business from the historical business processing information; The specific business content information is determined based on the business processing time information; Based on the specific business content information, a product business risk warning work order is generated, and user complaint warnings are issued based on the product business risk warning work order.

5. The user complaint early warning method as described in claim 4, characterized in that, The steps of generating a product business risk warning work order based on the specific business content information, and issuing a user complaint warning based on the product business risk warning work order, include: The business type to which the complained business belongs is determined based on the specific business content information; Get the number of user complaints for the product / service corresponding to the business type within a preset time period under different business handling organizations; When the number of user complaints exceeds a first preset threshold, obtain the process design information corresponding to the product business. Based on the process design information, a product business risk warning work order is generated, and user complaint warnings are issued based on the product business risk warning work order.

6. The user complaint early warning method as described in claim 4, characterized in that, The business early warning work order also includes: channel business processing anomaly early warning work order; After the step of determining the specific business content information based on the business processing time information, the method further includes: The organization responsible for handling the complaint is determined based on the specific business content information. Obtain all user complaint information within a preset time period, and determine the number of user complaints corresponding to the business processing organization based on the user complaint information; When the number of user complaints exceeds a second preset threshold, it is determined that the business processing organization has a business processing anomaly, and a channel business processing anomaly warning work order is output. User complaint alerts are issued based on the abnormality warning work orders processed through the aforementioned channel business.

7. A user complaint early warning device, characterized in that, The user complaint early warning device includes: The information acquisition module is used to acquire user complaint information within a preset time period; The information classification module is used to input the user complaint information into a preset Naive Bayes classification model to obtain the complaint classification result; The business matching module is used to determine the business being complained about based on the complaint classification results, and to perform back-tracking matching on the business being complained about through a preset business ordering big data model to obtain the historical business processing information of the business being complained about. The preset business ordering big data model is a business ordering model built for each user and contains the user's historical business processing records. The complaint warning module is used to generate a business warning work order based on the historical business processing information, and to issue a user complaint warning based on the business warning work order; The user complaint early warning device further includes a model training module, used to acquire historical user complaint information and extract keywords of a preset dimension from the historical user complaint information; to perform feature attribute segmentation on the keywords and construct model training samples based on the segmentation results, wherein the feature attribute is the category to which each keyword belongs; to input the model training samples into an initial Naive Bayes classification model for training, thereby obtaining multiple generations of Naive Bayes classification models to be tuned; and to determine the frequency of occurrence of the category calculated by each generation of Naive Bayes classification model in the training samples and the conditional probability of each feature attribute segmentation for each category, and to use the frequency of occurrence and the conditional probability as the model training results; and to perform parameter tuning on each generation of Naive Bayes classification model based on the model training results until the model converges, thereby obtaining a preset Naive Bayes classification model.

8. A user complaint early warning device, characterized in that, The device includes: a memory, a processor, and a user complaint warning program stored in the memory and executable on the processor, the user complaint warning program being configured to implement the steps of the user complaint warning method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium stores a user complaint warning program, which, when executed by a processor, implements the steps of the user complaint warning method as described in any one of claims 1 to 6.

Images