Complaint handling method, device, system and computer-readable storage medium

Abstract

The present invention discloses a complaint handling method, device, system and computer-readable storage medium, the method comprising: obtaining user feedback information, and inputting the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data; inputting the pre-complaint work order data into a pre-created judgment model set to determine target complaint work order data; inputting the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data; and generating a complaint statistical report based on the target complaint work order data and the complaint classification result. The present invention determines the pre-complaint work order data through a conversion processing model to ensure the content integrity of the pre-complaint work order data, and generates a complaint statistical report to conduct a comprehensive analysis of all complaints, thereby improving the accuracy of complaint handling, determining the target complaint work order data through a judgment model set, and determining the complaint classification result through a classification model, thereby improving the efficiency of complaint handling.

Description

Technical Field

[0001] The present invention relates to the field of financial technology, and in particular to complaint handling methods, devices, systems and computer-readable storage media. Background Art

[0002] With growing awareness of consumer rights protection, banks need to properly handle customer complaints, improve service levels and customer experience, and safeguard consumers' legitimate rights and interests. The current complaint handling process is as follows: users call the dedicated complaint hotline to file a complaint, and the operator manually records the complaint ticket. The complaint ticket and the transcript of the complaint recording are then passed to the problem-solving team, who manually analyze and confirm the complaint content before transferring it to the responsible unit for follow-up processing. This manual processing process is inefficient and time-consuming. Furthermore, manual recording of complaint tickets is highly subjective and prone to missing key information, which hinders the handler's understanding of the core needs of the user's complaint. Furthermore, the separation of user data within the data warehouse makes aggregation and processing difficult, making it difficult to conduct a comprehensive analysis of all complaints, resulting in low accuracy in complaint handling analysis.

[0003] Therefore, how to improve the efficiency and accuracy of complaint handling is an issue that needs to be addressed urgently. Summary of the Invention

[0004] The main purpose of the present invention is to provide a complaint handling method, device, system and computer-readable storage medium, aiming to solve the problem of how to improve the efficiency and accuracy of complaint handling.

[0005] To achieve the above objectives, the present invention provides a complaint handling method, which includes the following steps:

[0006] Obtain user feedback information and input the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data;

[0007] Input the pre-complaint work order data into a pre-created judgment model set to determine the target complaint work order data;

[0008] Input the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data;

[0009] Generate a complaint statistics report based on the target complaint work order data and the complaint classification results.

[0010] Optionally, the step of inputting the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data includes:

[0011] Inputting the user feedback information into a pre-created conversion processing model to obtain pre-complaint text data;

[0012] Obtain manually recorded pre-complaint work order data corresponding to the user feedback information, and determine the pre-complaint work order data based on the pre-complaint text data and the manually recorded pre-complaint work order data.

[0013] Optionally, the step of inputting the pre-complaint work order data into a pre-created judgment model set to determine the target complaint work order data includes:

[0014] Inputting the pre-complaint work order data into a pre-created judgment model set, and determining the non-complaint work order data and the first complaint work order data in the pre-complaint work order data by using the judgment model set;

[0015] Obtaining manual complaint classification results corresponding to the non-complaint work order data;

[0016] If the manual complaint classification result is that the category of the non-complaint work order data is complaint, determining that the non-complaint work order data is second complaint work order data, and determining target complaint work order data based on the first complaint work order data and the second complaint work order data;

[0017] If the manual complaint classification result is that the category of the non-complaint work order data is non-complaint, target complaint work order data is determined based on the first complaint work order data.

[0018] Optionally, the step of inputting the pre-complaint work order data into a pre-created judgment model set, and determining the non-complaint work order data and the first complaint work order data in the pre-complaint work order data by using the judgment model set includes:

[0019] Inputting the pre-complaint work order data into each judgment model in the pre-created judgment model set respectively, and obtaining the judgment result of each judgment model on the pre-complaint work order data;

[0020] The judgment result is corrected, and the non-complaint work order data and the first complaint work order data in the pre-complaint work order data are determined based on the corrected judgment result.

[0021] Optionally, the step of determining target complaint work order data according to the first complaint work order data and the second complaint work order data includes:

[0022] User status information corresponding to the first complaint work order data and the second complaint work order data is obtained respectively, and target complaint work order data is determined according to the first complaint work order data, the second complaint work order data and the user status information.

[0023] Optionally, the step of inputting the target complaint work order data into a pre-created classification model and determining a complaint classification result corresponding to the target complaint work order data includes:

[0024] Input the target complaint work order data into a pre-created classification model to determine the complaint pre-classification result corresponding to the target complaint work order data;

[0025] According to preset rules, the complaint pre-classification result is corrected to determine the complaint classification result corresponding to the target complaint work order data.

[0026] Optionally, if the manual complaint classification result is that the category of the non-complaint work order data is complaint, then after the step of determining that the non-complaint work order data is second complaint work order data, and determining the target complaint work order data based on the first complaint work order data and the second complaint work order data, the following steps are included:

[0027] Generate a training data set based on the second complaint work order data;

[0028] Each judgment model in the judgment model set is trained based on the training data set.

[0029] In addition, to achieve the above-mentioned purpose, the present invention further provides a complaint handling device, comprising:

[0030] An acquisition module is used to obtain user feedback information and input the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data;

[0031] A first determination module is configured to input the pre-complaint work order data into a pre-created judgment model set to determine target complaint work order data;

[0032] A second determination module is configured to input the target complaint work order data into a pre-created classification model to determine a complaint classification result corresponding to the target complaint work order data;

[0033] A generation module is used to generate a complaint statistical report based on the target complaint work order data and the complaint classification results.

[0034] Furthermore, the acquisition module further includes a determination module, and the determination module is configured to:

[0035] Inputting the user feedback information into a pre-created conversion processing model to obtain pre-complaint text data;

[0036] Obtain manually recorded pre-complaint work order data corresponding to the user feedback information, and determine the pre-complaint work order data based on the pre-complaint text data and the manually recorded pre-complaint work order data.

[0037] Furthermore, the first determining module is further configured to:

[0038] Inputting the pre-complaint work order data into a pre-created judgment model set, and determining the non-complaint work order data and the first complaint work order data in the pre-complaint work order data by using the judgment model set;

[0039] Obtaining manual complaint classification results corresponding to the non-complaint work order data;

[0040] If the manual complaint classification result is that the category of the non-complaint work order data is complaint, determining that the non-complaint work order data is second complaint work order data, and determining target complaint work order data based on the first complaint work order data and the second complaint work order data;

[0041] If the manual complaint classification result is that the category of the non-complaint work order data is non-complaint, target complaint work order data is determined based on the first complaint work order data.

[0042] Furthermore, the first determining module is further configured to:

[0043] Inputting the pre-complaint work order data into each judgment model in the pre-created judgment model set respectively, and obtaining the judgment result of each judgment model on the pre-complaint work order data;

[0044] The judgment result is corrected, and the non-complaint work order data and the first complaint work order data in the pre-complaint work order data are determined based on the corrected judgment result.

[0045] Furthermore, the first determining module is further configured to:

[0046] User status information corresponding to the first complaint work order data and the second complaint work order data is obtained respectively, and target complaint work order data is determined according to the first complaint work order data, the second complaint work order data and the user status information.

[0047] Furthermore, the second determining module is further configured to:

[0048] Input the target complaint work order data into a pre-created classification model to determine the complaint pre-classification result corresponding to the target complaint work order data;

[0049] According to preset rules, the complaint pre-classification result is corrected to determine the complaint classification result corresponding to the target complaint work order data.

[0050] Furthermore, the first determining module further includes a training module, and the training module is configured to:

[0051] Generate a training data set based on the second complaint work order data;

[0052] Each judgment model in the judgment model set is trained based on the training data set.

[0053] In addition, to achieve the above-mentioned purpose, the present invention also provides a complaint handling system, which includes: a memory, a processor, and a complaint handling program stored in the memory and executable on the processor, wherein the complaint handling program implements the steps of the complaint handling method described above when executed by the processor.

[0054] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which a complaint handling program is stored. When the complaint handling program is executed by a processor, the steps of the complaint handling method described above are implemented.

[0055] The complaint handling method proposed in the present invention obtains user feedback information and inputs the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data; inputs the pre-complaint work order data into a pre-created judgment model set to determine target complaint work order data; inputs the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data; and generates a complaint statistical report based on the target complaint work order data and the complaint classification result. The present invention determines the pre-complaint work order data through the conversion processing model to ensure the content integrity of the pre-complaint work order data, and generates a complaint statistical report to conduct a comprehensive analysis of all complaints, thereby improving the accuracy of complaint handling, determining the target complaint work order data through the judgment model set, and determining the complaint classification result through the classification model, thereby improving the efficiency of complaint handling. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention;

[0057] Figure 2 This is a flowchart of a first embodiment of a complaint handling method according to the present invention;

[0058] Figure 3 This is a flow chart of a second embodiment of the complaint handling method of the present invention;

[0059] Figure 4 This is a flowchart of a third embodiment of the complaint handling method of the present invention;

[0060] Figure 5 2 is a flow chart of a fourth embodiment of the complaint handling method of the present invention.

[0061] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and embodiments. DETAILED DESCRIPTION

[0062] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0063] like Figure 1 As shown, Figure 1 It is a schematic diagram of the device structure of the hardware operating environment involved in the embodiment of the present invention.

[0064] The device in the embodiment of the present invention may be a PC or a server device.

[0065] like Figure 1 As shown, the device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, and a communication bus 1002. The communication bus 1002 is used to implement communication between these components. The user interface 1003 may include a display and an input unit such as a keyboard. Optionally, the user interface 1003 may also include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory, such as a disk storage device. The memory 1005 may also be a storage device independent of the processor 1001.

[0066] Those skilled in the art will understand that Figure 1 The device structure shown in the figure does not constitute a limitation of the device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0067] like Figure 1 As shown, the memory 1005 as a computer storage medium may include an operating system, a network communication module, a user interface module and a complaint handling program.

[0068] Among them, the operating system is a program that manages and controls the portable complaint handling system and software resources, and supports the operation of the network communication module, user interface module, complaint handling program and other programs or software; the network communication module is used to manage and control the network interface 1004; the user interface module is used to manage and control the user interface 1003.

[0069] exist Figure 1 In the complaint handling system shown, the complaint handling system calls the complaint handling program stored in the memory 1005 through the processor 1001 and executes the operations in each embodiment of the complaint handling method described below.

[0070] Based on the above hardware structure, an embodiment of the complaint handling method of the present invention is proposed.

[0071] Reference Figure 2 , Figure 2 This is a flowchart of a first embodiment of a complaint handling method according to the present invention, wherein the method comprises:

[0072] Step S10: Obtain user feedback information, and input the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data;

[0073] Step S20: input the pre-complaint work order data into a pre-created judgment model set to determine the target complaint work order data;

[0074] Step S30: input the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data;

[0075] Step S40: Generate a complaint statistics report based on the target complaint work order data and the complaint classification result.

[0076] The complaint handling method of this embodiment is applied to the complaint handling system of a financial institution. The complaint handling system can be applied to a terminal or PC device. For ease of description, the complaint handling system is used as an example for description. The complaint handling system obtains user feedback information, inputs the user feedback information into a pre-created conversion processing model, and obtains pre-complaint text data. The manually recorded pre-complaint work order data corresponding to the user feedback information is obtained, and the pre-complaint work order data is determined based on the pre-complaint text data and the manually recorded pre-complaint work order data. The complaint handling system inputs the pre-complaint work order data into a pre-created judgment model set to determine the target complaint work order data. The complaint handling system inputs the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data. The complaint handling system generates a complaint statistical report based on the target complaint work order data and the complaint classification result.

[0077] The complaint handling method of this embodiment obtains user feedback information and inputs the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data; inputs the pre-complaint work order data into a pre-created judgment model set to determine target complaint work order data; inputs the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data; and generates a complaint statistical report based on the target complaint work order data and the complaint classification result. The present invention determines the pre-complaint work order data through a conversion processing model to ensure the content integrity of the pre-complaint work order data, and generates a complaint statistical report to conduct a comprehensive analysis of all complaints, thereby improving the accuracy of complaint handling, determining the target complaint work order data through a judgment model set, and determining the complaint classification result through a classification model, thereby improving the efficiency of complaint handling.

[0078] The following describes each step in detail:

[0079] Step S10: Obtain user feedback information, and input the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data;

[0080] In this embodiment, the complaint handling system obtains user feedback information and inputs the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data; it can be understood that user feedback information can be voice information that users feedback by calling customer service, or text information feedback through channels such as mini-programs and apps. The pre-created conversion processing model can convert and process the voice information, text information and other information fed back by users into pre-complaint work order information that is recognizable, structured, and complete in content by the complaint handling system, so as to ensure that the user's feedback information is fully recorded, avoid subjective recording of complaint content, and avoid omitting key information in the user's feedback information that affects the understanding of the core demands of the user's complaint, thereby improving the accuracy of complaint handling.

[0081] Specifically, step S10 includes:

[0082] Step S101: inputting the user feedback information into a pre-created conversion processing model to obtain pre-complaint text data;

[0083] In this step, after obtaining user feedback information, the complaint processing system inputs the user feedback information into a pre-created conversion processing model, extracts key information from the user feedback information through the conversion processing model, and obtains pre-complaint text data.

[0084] Exemplarily, the pre-created conversion processing model includes an ASR (Automatic Speech Recognition) speech translation model. User feedback information is the voice information provided by users when they call customer service, i.e., the phone recording. The complaint handling system inputs the phone recording into the ASR speech translation model and converts the phone recording into pre-complaint text data. Optionally, the pre-created conversion processing model may also include a text recognition model. When the complaint handling system receives user feedback information provided through channels such as mini-programs and apps, the text recognition model converts the corresponding text information into pre-complaint text data.

[0085] Step S201: Obtain manually recorded pre-complaint work order data corresponding to the user feedback information, and determine the pre-complaint work order data based on the pre-complaint text data and the manually recorded pre-complaint work order data.

[0086] In this step, when the user provides corresponding user feedback information through channels such as calling customer service telephones, mini-programs, and apps, the relevant customer service personnel will manually process the received user feedback information, obtain the corresponding manually recorded pre-complaint work order data based on the user feedback information, and upload the manually recorded pre-complaint work order data to the complaint processing system. The complaint processing system merges the manually recorded pre-complaint work order data and the pre-complaint text data determined by the conversion processing model. After data cleaning and conversion, the merged data is obtained to obtain structured pre-complaint work order data with a unified format. It is understandable that the manually recorded pre-complaint work order data obtained by manual processing by the relevant customer service personnel may contain subjective recording of complaint content, omission of key information, etc., which may affect the understanding of the core demands of the complaint. Therefore, the pre-complaint work order data is determined in combination with the pre-complaint text data determined by the conversion processing model to avoid subjective recording of complaint content and omission of key information in the user's feedback information, which may affect the understanding of the core demands of the user's complaint and improve the accuracy of complaint processing. That is, the main purpose of the pre-complaint text data determined by the conversion processing model is to supplement the manually entered complaint work order with complete complaint content.

[0087] Step S20: input the pre-complaint work order data into a pre-created judgment model set to determine the target complaint work order data;

[0088] In this embodiment, after determining the pre-complaint work order data, the complaint handling system inputs the pre-complaint work order data into a pre-created judgment model set, and determines the target complaint work order data in the pre-complaint work order data through the judgment model set; it should be noted that there may be work order data in the pre-complaint work order data that is actually not a complaint, so a judgment model set is required to determine the target complaint work order data in the pre-complaint work order data.

[0089] Step S30: input the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data;

[0090] In this embodiment, the complaint handling system inputs the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data. Exemplarily, the classification model is a TextCNN classification model, which includes complaint channel, complaint business, and complaint reason classification. The complaint handling system classifies the target complaint work order data through the classification model to determine the complaint classification result corresponding to the target complaint work order data, including: complaint channel, complaint business, and complaint reason.

[0091] Step S40: Generate a complaint statistics report based on the target complaint work order data and the complaint classification result.

[0092] In this embodiment, the complaint handling system generates complaint statistical reports based on the target complaint work order data and complaint classification results. Specifically, the complaint handling system performs statistical analysis on the data of each process from obtaining user feedback information to determining the complaint classification results, and finally outputs complaint statistical reports of different dimensions for monitoring and management, so as to conduct all-weather, full-process, and real-time monitoring and management of complaints.

[0093] The complaint handling system of this embodiment obtains user feedback information and inputs the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data; it can be understood that user feedback information can be voice information that users feedback by calling customer service, or text information feedback through channels such as mini-programs and apps. The pre-created conversion processing model can convert and process the voice information, text information and other information fed back by users into pre-complaint work order information that is recognizable, structured, and complete in content by the complaint handling system, so as to ensure that the user's feedback information is fully recorded, avoid subjective recording of complaint content, and avoid omitting key information in the user's feedback information that affects the understanding of the core demands of the user's complaint, thereby improving the accuracy of complaint handling.

[0094] Further, refer to Figure 3 Based on the first embodiment of the complaint handling method of the present invention, a second embodiment of the complaint handling method of the present invention is proposed.

[0095] The second embodiment of the complaint handling method differs from the first embodiment in that step S20 includes:

[0096] Step S201: input the pre-complaint work order data into a pre-created judgment model set, and determine the non-complaint work order data and the first complaint work order data in the pre-complaint work order data by using the judgment model set;

[0097] In this step, after determining the pre-complaint work order data, the complaint handling system first corrects the accuracy of the pre-complaint work order data based on the NLP algorithm model, and then inputs the corrected pre-complaint work order data into a pre-created judgment model set, and determines the non-complaint work order data and the first complaint work order data in the pre-complaint work order data through the judgment model set; it should be noted that the judgment model set includes multiple types of judgment models, forming a multi-model fusion, and determining the non-complaint work order data and the first complaint work order data based on multiple types of judgment models, thereby improving the accuracy of complaint judgment and thereby improving the accuracy of complaint handling.

[0098] Specifically, step S201 includes:

[0099] Step S2011: input the pre-complaint work order data into each judgment model in the pre-created judgment model set, and obtain the judgment result of each judgment model on the pre-complaint work order data;

[0100] In this step, the complaint handling system inputs the corrected pre-complaint work order data into each judgment model in the pre-created judgment model set, and obtains the judgment result of each judgment model on the pre-complaint work order data; in a feasible embodiment, the judgment model set includes a FastText-based judgment model, a HAN-based judgment model, a TextCNN-based judgment model, and a DPCNN-based judgment model, forming a multi-model fusion, and each of the above models judges the pre-complaint work order data to obtain the corresponding judgment result.

[0101] Optionally, before inputting the pre-complaint work order data into each judgment model in the pre-created judgment model set, the complaint handling system uses the pre-rules to predict the pre-complaint work order data, predicts and determines the non-complaint work order data in the pre-complaint work order data, sends the non-complaint work order data to the corresponding customer service personnel for review, and removes the non-complaint work order data from the pre-complaint work order data, and then inputs the remaining pre-complaint work order data into each judgment model in the pre-created judgment model set. It can be understood that the pre-complaint work order data is first predicted to filter out some non-complaint work order data, thereby reducing the amount of calculation for subsequent calculations and improving the efficiency of complaint handling, and the pre-rules are used to predict and filter out some non-complaint work order data, and then the pre-rules are used to predict and filter out the rest, and multiple screenings are performed to avoid missing non-complaint work order data and improve the accuracy of subsequent complaint handling.

[0102] Step S2012: correct the judgment result, and determine the non-complaint work order data and the first complaint work order data in the pre-complaint work order data based on the corrected judgment result.

[0103] In this step, after obtaining the judgment result of each judgment model on the pre-complaint work order data, the complaint handling system corrects each judgment result, and fuses the corrected judgment results to obtain a target judgment result, and determines the non-complaint work order data and the first complaint work order data in the pre-complaint work order data according to the target judgment result; optionally, the complaint handling system fuses and analyzes the judgment result of each judgment model on the pre-complaint work order data to obtain a fused judgment result, and then corrects the fused judgment result to obtain a target judgment result, and determines the non-complaint work order data and the first complaint work order data in the pre-complaint work order data according to the target judgment result; in a feasible embodiment, the judgment model set It includes a judgment model based on FastText, a judgment model based on HAN, a judgment model based on TextCNN, and a judgment model based on DPCNN. Each of the above judgment models judges the pre-complaint work order to obtain a judgment result. The complaint processing system integrates and analyzes each judgment result to obtain a fused judgment result, and then corrects the fused judgment result through the three classification models of complaint channel, complaint business, and complaint reason in the HAN judgment model to obtain a target judgment result. The pre-complaint work order data determined by the target judgment result to be a complaint is used as the first complaint work order data, and the pre-complaint work order data determined by the target judgment result to be a non-complaint work order data is used as the non-complaint work order data.

[0104] Step S202: Obtain the manual complaint classification result corresponding to the non-complaint work order data;

[0105] In this step, after the complaint handling system determines the non-complaint work orders in the pre-complaint work order data, it will send the non-complaint work order data to the corresponding customer service personnel for manual review. After the relevant customer service personnel meet the requirements, they will obtain the corresponding manual complaint classification results, and then upload the corresponding manual complaint classification results to the complaint handling system. The complaint handling system obtains the manual complaint classification results corresponding to the non-complaint work order data and analyzes the manual complaint classification results corresponding to the non-complaint work order data.

[0106] Step S203: If the manual complaint classification result is that the category of the non-complaint work order data is complaint, the non-complaint work order data is determined to be second complaint work order data, and target complaint work order data is determined based on the first complaint work order data and the second complaint work order data;

[0107] In this step, if the complaint handling system determines that the manual complaint classification result is that the category of the non-complaint work order data is complaint, the non-complaint work order data is determined to be the second complaint work order data, and the target complaint work order data is determined based on the first complaint work order data and the second complaint work order data.

[0108] Furthermore, step S203 includes:

[0109] Step S2031: Obtain user status information corresponding to the first complaint work order data and the second complaint work order data respectively, and determine target complaint work order data based on the first complaint work order data, the second complaint work order data and the user status information.

[0110] In this step, after determining the first complaint work order data and the second complaint work order data, the complaint handling system obtains the user status information corresponding to the first complaint work order data and the second complaint work order data respectively. The user status information includes customer portrait, transaction information, sentiment index and other information. The target complaint work order data is determined by combining the first complaint work order data, the second complaint work order data and the user status information. It can be understood that the first complaint work order data and the second complaint work order data usually contain multiple complaint work order data. The complaint handling system obtains the user status information corresponding to each complaint work order data, and then combines each complaint work order data and the corresponding user status information to obtain the target complaint work order data corresponding to each complaint work order data.

[0111] Step S204: If the manual complaint classification result is that the category of the non-complaint work order data is non-complaint, target complaint work order data is determined based on the first complaint work order data.

[0112] In this step, if the complaint handling system determines that the manual complaint classification result is non-complaint and the category of the work order data is non-complaint, it obtains the user status information corresponding to the first complaint work order data. The user status information includes customer portrait, transaction information, sentiment index and other information. The target complaint work order data is determined by combining the first complaint work order data and the user status information.

[0113] The complaint handling system of this embodiment inputs pre-complaint work order data into a pre-created set of judgment models, and uses the judgment model set to determine non-complaint work order data and first complaint work order data within the pre-complaint work order data. The system then obtains the manual complaint classification results corresponding to the non-complaint work order data. If the manual complaint classification result indicates that the non-complaint work order data is classified as complaint, the non-complaint work order data is determined to be second complaint work order data, and target complaint work order data is determined based on the first and second complaint work order data. If the manual complaint classification result indicates that the non-complaint work order data is classified as non-complaint, the target complaint work order data is determined based on the first complaint work order data. Determining target complaint work order data through the judgment model set improves the efficiency of complaint handling.

[0114] Further, refer to Figure 4 Based on the first and second embodiments of the complaint handling method of the present invention, a third embodiment of the complaint handling method of the present invention is proposed.

[0115] The third embodiment of the complaint handling method differs from the first and second embodiments in that step S30 includes:

[0116] Step S301: input the target complaint work order data into a pre-created classification model to determine the complaint pre-classification result corresponding to the target complaint work order data;

[0117] Step S302: According to preset rules, the complaint pre-classification result is corrected to determine the complaint classification result corresponding to the target complaint work order data.

[0118] In this embodiment, the complaint handling system inputs the target complaint work order data into a pre-created classification model, determines the complaint pre-classification result corresponding to the target complaint work order data, and performs correction operations on the complaint pre-classification result according to preset rules to determine the complaint classification result corresponding to the target complaint work order data; in a feasible embodiment, after determining the target complaint work order data, the complaint handling system performs three-level complaint classification on the target complaint work order data, and uses the complaint channel, complaint business, and complaint motivation classification model based on TextCNN to determine the complaint classification result corresponding to the target complaint work order data, including: complaint channel, complaint business, and complaint reason. After determining the complaint classification result, the preset post-rule is used to correct the complaint classification result. After correction, the complaint classification result corresponding to the target complaint work order data includes: complaint channel, complaint business, and complaint reason.

[0119] In this embodiment, the complaint handling system inputs target complaint ticket data into a pre-created classification model, determines the pre-classification results for the target complaint ticket data, and then, based on pre-set rules, modifies the pre-classification results to determine the corresponding complaint classification results. By modifying the classification results for the target complaint ticket data determined by the classification model, the accuracy of the complaint classification results is ensured, thereby improving the accuracy of complaint handling.

[0120] Further, refer to Figure 5 Based on the first to third embodiments of the complaint handling method of the present invention, a fourth embodiment of the complaint handling method of the present invention is proposed.

[0121] The fourth embodiment of the complaint handling method differs from the first to third embodiments of the complaint handling method in that, if the manual complaint classification result indicates that the category of the non-complaint work order data is complaint, then after the steps of determining that the non-complaint work order data is second complaint work order data and determining target complaint work order data based on the first complaint work order data and the second complaint work order data, the method further includes:

[0122] Step a: generating a training data set based on the second complaint work order data;

[0123] Step b: training each judgment model in the judgment model set based on the training data set.

[0124] In this embodiment, when the complaint handling system determines the non-complaint work order data through the judgment model and the relevant customer service personnel determine that the category of the non-complaint work order data is complaint, the complaint handling system determines the non-complaint work order data as second complaint work order data, and determines that the judgment model set has misjudged this type of complaint. Therefore, the complaint handling system generates a training data set for the second complaint work order data, and trains each judgment model in the judgment model set based on the training data set, so that the judgment model set can improve the accuracy of judging such complaints.

[0125] The complaint handling system of this embodiment generates a training data set based on the second complaint work order data, and trains each judgment model in the judgment model set based on the training data set, thereby improving the accuracy of the judgment model set in complaint judgment, and thereby improving the accuracy of complaint handling.

[0126] The present invention also provides a complaint handling device, which includes:

[0127] An acquisition module is used to obtain user feedback information and input the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data;

[0128] A first determination module is configured to input the pre-complaint work order data into a pre-created judgment model set to determine target complaint work order data;

[0129] A second determination module is configured to input the target complaint work order data into a pre-created classification model to determine a complaint classification result corresponding to the target complaint work order data;

[0130] A generation module is used to generate a complaint statistical report based on the target complaint work order data and the complaint classification results.

[0131] Furthermore, the acquisition module further includes a determination module, and the determination module is configured to:

[0132] Inputting the user feedback information into a pre-created conversion processing model to obtain pre-complaint text data;

[0133] Obtain manually recorded pre-complaint work order data corresponding to the user feedback information, and determine the pre-complaint work order data based on the pre-complaint text data and the manually recorded pre-complaint work order data.

[0134] Furthermore, the first determining module is further configured to:

[0135] Inputting the pre-complaint work order data into a pre-created judgment model set, and determining the non-complaint work order data and the first complaint work order data in the pre-complaint work order data by using the judgment model set;

[0136] Obtaining manual complaint classification results corresponding to the non-complaint work order data;

[0137] If the manual complaint classification result is that the category of the non-complaint work order data is complaint, determining that the non-complaint work order data is second complaint work order data, and determining target complaint work order data based on the first complaint work order data and the second complaint work order data;

[0138] If the manual complaint classification result is that the category of the non-complaint work order data is non-complaint, target complaint work order data is determined based on the first complaint work order data.

[0139] Furthermore, the first determining module is further configured to:

[0140] Inputting the pre-complaint work order data into each judgment model in the pre-created judgment model set respectively, and obtaining the judgment result of each judgment model on the pre-complaint work order data;

[0141] The judgment result is corrected, and the non-complaint work order data and the first complaint work order data in the pre-complaint work order data are determined based on the corrected judgment result.

[0142] Furthermore, the first determining module is further configured to:

[0143] User status information corresponding to the first complaint work order data and the second complaint work order data is obtained respectively, and target complaint work order data is determined according to the first complaint work order data, the second complaint work order data and the user status information.

[0144] Furthermore, the second determining module is further configured to:

[0145] Input the target complaint work order data into a pre-created classification model to determine the complaint pre-classification result corresponding to the target complaint work order data;

[0146] According to preset rules, the complaint pre-classification result is corrected to determine the complaint classification result corresponding to the target complaint work order data.

[0147] Furthermore, the first determining module further includes a training module, and the training module is configured to:

[0148] Generate a training data set based on the second complaint work order data;

[0149] Each judgment model in the judgment model set is trained based on the training data set.

[0150] The invention also provides a complaint handling system.

[0151] The complaint handling system of the present invention includes: a memory, a processor, and a complaint handling program stored in the memory and executable on the processor. When the complaint handling program is executed by the processor, the steps of the complaint handling method described above are implemented.

[0152] Among them, the method implemented when the complaint handling program running on the processor is executed can refer to the various embodiments of the complaint handling method of the present invention, and will not be repeated here.

[0153] The present invention also provides a computer-readable storage medium.

[0154] The computer-readable storage medium of the present invention stores a complaint handling program, which implements the steps of the complaint handling method described above when executed by a processor.

[0155] Among them, the method implemented when the complaint handling program running on the processor is executed can refer to the various embodiments of the complaint handling method of the present invention, and will not be repeated here.

[0156] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or system comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or system. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or system comprising the element.

[0157] The serial numbers of the above embodiments of the present invention are for description only and do not represent the advantages or disadvantages of the embodiments.

[0158] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better embodiment. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0159] The above are only preferred embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for handling complaints, characterized in that: The complaint handling method includes the following steps: Obtain user feedback information and input the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data; Inputting the pre-complaint work order data into a pre-created judgment model set to determine the target complaint work order data, wherein the judgment model set includes a judgment model based on FastText, a judgment model based on HAN, a judgment model based on TextCNN, and a judgment model based on DPCNN, forming a multi-model fusion; Input the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data; Generate a complaint statistics report based on the target complaint work order data and the complaint classification results; The step of inputting the pre-complaint work order data into a pre-created judgment model set includes: The pre-complaint work order data is predicted using the pre-rules to predict the non-complaint work order data in the pre-complaint work order data, the non-complaint work order data is sent to the customer service staff for review, and after the non-complaint work order data is removed from the pre-complaint work order data, the remaining pre-complaint work order data is input into each judgment model in the pre-created judgment model set.

2. The complaint handling method according to claim 1, wherein: The step of inputting the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data includes: Inputting the user feedback information into a pre-created conversion processing model to obtain pre-complaint text data; Obtain manually recorded pre-complaint work order data corresponding to the user feedback information, and determine the pre-complaint work order data based on the pre-complaint text data and the manually recorded pre-complaint work order data.

3. The complaint handling method according to claim 1, wherein: The step of inputting the pre-complaint work order data into a pre-created judgment model set to determine the target complaint work order data includes: Inputting the pre-complaint work order data into a pre-created judgment model set, and determining the non-complaint work order data and the first complaint work order data in the pre-complaint work order data by using the judgment model set; Obtaining manual complaint classification results corresponding to the non-complaint work order data; If the manual complaint classification result is that the category of the non-complaint work order data is complaint, determining that the non-complaint work order data is second complaint work order data, and determining target complaint work order data based on the first complaint work order data and the second complaint work order data; If the manual complaint classification result is that the category of the non-complaint work order data is non-complaint, target complaint work order data is determined based on the first complaint work order data.

4. The complaint handling method according to claim 3, wherein: The step of inputting the pre-complaint work order data into a pre-created judgment model set and determining the non-complaint work order data and the first complaint work order data in the pre-complaint work order data by using the judgment model set includes: Inputting the pre-complaint work order data into each judgment model in the pre-created judgment model set respectively, and obtaining the judgment result of each judgment model on the pre-complaint work order data; The judgment result is corrected, and the non-complaint work order data and the first complaint work order data in the pre-complaint work order data are determined based on the corrected judgment result.

5. The complaint handling method according to claim 3, wherein: The step of determining target complaint work order data based on the first complaint work order data and the second complaint work order data includes: User status information corresponding to the first complaint work order data and the second complaint work order data is obtained respectively, and target complaint work order data is determined according to the first complaint work order data, the second complaint work order data and the user status information.

6. The complaint handling method according to claim 1, wherein: The step of inputting the target complaint work order data into a pre-created classification model to determine the complaint classification result corresponding to the target complaint work order data includes: Input the target complaint work order data into a pre-created classification model to determine the complaint pre-classification result corresponding to the target complaint work order data; According to preset rules, the complaint pre-classification result is corrected to determine the complaint classification result corresponding to the target complaint work order data.

7. The complaint handling method according to claim 3, wherein: If the manual complaint classification result is that the category of the non-complaint work order data is complaint, then determining the non-complaint work order data as second complaint work order data, and determining target complaint work order data based on the first complaint work order data and the second complaint work order data, the method further includes: Generate a training data set based on the second complaint work order data; Each judgment model in the judgment model set is trained based on the training data set.

8. A complaint handling device, characterized in that: The complaint handling device comprises: An acquisition module is used to obtain user feedback information and input the user feedback information into a pre-created conversion processing model to determine pre-complaint work order data; A first determination module is configured to input the pre-complaint work order data into a pre-created judgment model set to determine target complaint work order data, wherein the judgment model set includes a FastText-based judgment model, a HAN-based judgment model, a TextCNN-based judgment model, and a DPCNN-based judgment model to form a multi-model fusion; A second determination module is configured to input the target complaint work order data into a pre-created classification model to determine a complaint classification result corresponding to the target complaint work order data; A generation module, configured to generate a complaint statistical report based on the target complaint work order data and the complaint classification results; The first determination module is also used to predict the pre-complaint work order data using the pre-rules, predict and determine the non-complaint work order data in the pre-complaint work order data, send the non-complaint work order data to the customer service personnel for review, and after removing the non-complaint work order data from the pre-complaint work order data, input the remaining pre-complaint work order data into each judgment model in the pre-created judgment model set.

9. A complaint handling system, characterized in that: The complaint handling system includes: a memory, a processor, and a complaint handling program stored in the memory and executable on the processor. When the complaint handling program is executed by the processor, the steps of the complaint handling method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a complaint handling program, which, when executed by a processor, implements the steps of the complaint handling method according to any one of claims 1 to 7.

Images