Feedback information processing method and device, computer equipment and readable storage medium

Abstract

The invention discloses a feedback information processing method and device, computer equipment and a readable storage medium, relates to the field of data processing technology, medical treatment and health and the field of financial science and technology, can accurately understand the essence of customer feedback, avoids subjective deviation and understanding errors possibly occurring in manual analysis, and improves the processing efficiency. The method comprises the following steps: collecting feedback information of a plurality of clients for data preprocessing to obtain feedback information to be processed; based on a preset intelligent analysis module, determining an emotional tendency and demand content corresponding to each piece of to-be-processed feedback information, determining a feedback category and a processing priority according to the emotional tendency and the demand content, and sorting the plurality of pieces of to-be-processed feedback information according to the feedback category and the processing priority corresponding to each piece of to-be-processed feedback information; and based on the sorting result, determining the processing complexity corresponding to each piece of to-be-processed feedback information and a processing flow matched with the processing complexity, and processing the to-be-processed feedback information according to the processing flow.

Description

Technical Field

[0001] The present application relates to the fields of data processing technology and financial technology, and in particular to a feedback information processing method, apparatus, computer equipment and readable storage medium. Background Art

[0002] With the deep integration of FinTech and the health insurance industry, the intelligent and digital transformation of customer service has become a core focus for insurance institutions to enhance their competitiveness. Driven by FinTech, customer feedback data generated across health insurance business scenarios (such as insurance consultation, claims services, and policy management) is growing exponentially. This data, encompassing multiple modalities such as phone calls, online chat texts, social media comments, and images of medical receipts, places higher demands on data processing efficiency and analytical depth.

[0003] Among related technologies, applications in the customer service field are primarily focused on basic functions. For example, some institutions employ rule-based intelligent customer service systems that respond to customer inquiries through fixed keyword matching. Some institutions use statistical tools to perform simple classifications of feedback data (e.g., counting the number of responses by channel or issue type) and rely on the experience of customer service staff to determine issue priorities and solutions.

[0004] In the process of implementing this application, the applicant discovered that the related technology has at least the following problems:

[0005] In traditional customer service processes, the collection, classification, processing, and tracking of customer feedback rely primarily on manual labor. This fully manual process leads to inefficient feedback processing and delayed responses. Furthermore, due to varying levels of experience among customer service personnel, similar issues can be misclassified or inconsistently handled, severely impacting the stability of the customer service experience. Summary of the Invention

[0006] In light of this, this application provides a feedback information processing method, apparatus, computer device, and readable storage medium. The primary purpose is to address the issue of customer feedback collection, classification, processing, and tracking primarily relying on manual operations. This full-process manual intervention model results in inefficient feedback processing and untimely response to issues. Furthermore, due to varying levels of customer service personnel experience, similar issues can be misclassified or their processing results can be inconsistent, seriously impacting the stability of the customer service experience.

[0007] According to the first aspect of the present application, a method for processing feedback information is provided, the method comprising:

[0008] Collecting multiple customer feedback information, performing data preprocessing on the multiple customer feedback information to obtain multiple feedback information to be processed, and inputting the multiple feedback information to be processed into a preset intelligent analysis module, wherein each of the customer feedback information includes feedback data of at least one modality;

[0009] Determining, based on the preset intelligent analysis module, the emotional tendency and demand content corresponding to each of the feedback information to be processed, and determining the feedback category and processing priority corresponding to the corresponding feedback information to be processed according to the emotional tendency and demand content, and sorting the plurality of feedback information to be processed according to the feedback category and processing priority corresponding to each of the feedback information to be processed;

[0010] Based on the ranking results, determine the processing complexity corresponding to each piece of feedback information to be processed and the processing flow that matches the processing complexity according to the feedback knowledge graph in the knowledge base, and process the feedback information to be processed according to the processing flow;

[0011] Record the processing flow and processing results corresponding to each of the feedback information to be processed, and update the feedback knowledge graph in the knowledge base according to the processing flow and processing results corresponding to each of the feedback information to be processed.

[0012] According to a second aspect of the present application, a feedback information processing device is provided, the device comprising:

[0013] a collection module configured to collect a plurality of customer feedback information, perform data preprocessing on the plurality of customer feedback information to obtain a plurality of feedback information to be processed, and input the plurality of feedback information to be processed into a preset intelligent analysis module, wherein each of the customer feedback information includes feedback data of at least one modality;

[0014] an analysis module configured to determine, based on the preset intelligent analysis module, the emotional tendency and demand content corresponding to each piece of feedback information to be processed, determine the feedback category and processing priority corresponding to the corresponding piece of feedback information to be processed according to the emotional tendency and demand content, and sort the plurality of pieces of feedback information to be processed according to the feedback category and processing priority corresponding to each piece of feedback information to be processed;

[0015] A processing module is configured to determine, based on the sorting result and according to the feedback knowledge graph in the knowledge base, the processing complexity corresponding to each piece of feedback information to be processed, and a processing flow matching the processing complexity, and process the feedback information to be processed according to the processing flow;

[0016] The optimization module is used to record the processing flow and processing results corresponding to each of the feedback information to be processed, and update the feedback knowledge graph in the knowledge base according to the processing flow and processing results corresponding to each of the feedback information to be processed.

[0017] According to a third aspect of the present application, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the methods described in the first aspect when executing the computer program.

[0018] According to a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0019] By means of the above technical solutions, the present application provides a feedback information processing method, apparatus, computer device, and readable storage medium. The embodiments of the present application determine the emotional tendency and demand content of each pending feedback message through a preset intelligent analysis module, enabling a deeper and more accurate understanding of the essence of customer feedback, avoiding the subjective bias and misunderstanding that may occur in manual analysis. Furthermore, the embodiments of the present application sort multiple pending feedback messages according to feedback category and processing priority, allowing processing personnel to clearly understand the importance and priority of the feedback, avoiding confusion and disorder in feedback processing. Thus, when faced with a large amount of customer feedback, processing personnel can directly start processing from high-priority feedback, improving processing efficiency. Furthermore, the present application can adopt different processing methods for pending feedback messages of varying complexity. Simple feedback can be quickly processed using an automated process, while complex feedback can be processed by professionals for in-depth analysis and processing, improving the pertinence and effectiveness of the processing. Furthermore, the knowledge base provides customer service personnel with rich knowledge support, improving service efficiency. At the same time, the self-learning capability enables the system to continuously update the knowledge base based on new customer feedback and processing experience, keeping pace with the times and improving service quality and accuracy.

[0020] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0022] Figure 1 A schematic diagram of a feedback information processing method provided in an embodiment of the present application is shown;

[0023] Figure 2 A schematic diagram of another feedback information processing method provided in an embodiment of the present application is shown;

[0024] Figure 3 A schematic diagram of another feedback information processing method provided in an embodiment of the present application is shown;

[0025] Figure 4 A schematic diagram of another feedback information processing method provided in an embodiment of the present application is shown;

[0026] Figure 5 A schematic structural diagram of a feedback information processing device provided in an embodiment of the present application is shown;

[0027] Figure 6 A schematic diagram of the device structure of a computer device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0028] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and are not to be construed as limiting the present application.

[0029] Those skilled in the art will understand that, unless otherwise stated, the singular forms "a," "an," "said," and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of this application refers to the presence of the stated features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0030] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0031] Those skilled in the art will appreciate that the term "terminal" as used herein includes both devices that are wireless signal receivers, i.e., devices that only have wireless signal receivers without transmission capabilities, and devices that have receiving and transmitting hardware capable of performing two-way communication over a two-way communication link. Such devices may include: cellular or other communication devices with single-line displays, multi-line displays, or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service) devices that may combine voice, data processing, fax, and / or data communication capabilities; PDAs (Personal Digital Assistants) that may include a radio frequency receiver, a pager, Internet / Intranet access, a web browser, a notepad, a calendar, and / or a GPS (Global Positioning System) receiver; and conventional laptop and / or palmtop computers or other devices that have and / or include a radio frequency receiver. As used herein, a "terminal" may be portable, transportable, installed in a vehicle (air, sea, and / or land), or adapted and / or configured to operate locally, and / or in a distributed manner, at any other location on Earth and / or in space. As used herein, a "terminal" may also be a communication terminal, an Internet access terminal, or a music / video playback terminal, such as a PDA, an MID (Mobile Internet Device), and / or a mobile phone with music / video playback capabilities, or a device such as a smart TV or a set-top box.

[0032] This technical solution is applicable to the customer service and feedback optimization system of the health insurance industry. Through multi-module collaboration, the system realizes full-link closed-loop management from customer feedback collection to processing result feedback optimization, focusing on solving the problems of inefficient manual processing and insufficient data utilization in traditional services.

[0033] Customer feedback refers to service-related opinions or questions submitted by customers through multiple channels, such as phone, email, social media, and online chat. It includes feedback data in at least one modality, including but not limited to unstructured data such as text, voice, and image. For example, if a customer complains by phone about "slow claim processing," they will receive feedback in voice mode, or if they upload a screenshot of their insurance policy on social media with the comment "terms unclear," they will receive feedback in both text and image modes.

[0034] The intelligent analysis module is the core functional module of the system. It integrates deep learning, natural language processing and other technologies based on DeepSeek technology to identify the sentiment tendency and extract the demand content of the feedback information to be processed, and determines the feedback category and processing priority in combination with the knowledge base.

[0035] The feedback knowledge graph is a structured knowledge network stored in the knowledge base, organizing customer feedback-related knowledge in an entity-relationship format. Entities include customer entities, such as "female customer aged 30-40," problem entities, such as "slow claim processing," processing entities, such as "transferred to expert customer service," and result entities, such as "customer satisfaction." Relationships include "problem-processing" and "processing-result" associations, such as "slow claim processing - transferred to expert customer service - customer satisfaction." The knowledge graph is continuously updated through self-learning.

[0036] The embodiment of the present application provides a method for processing feedback information, such as Figure 1 As shown, the method includes:

[0037] S10. Collect multiple customer feedback information, perform data preprocessing on the multiple customer feedback information to obtain multiple feedback information to be processed, and input the multiple feedback information to be processed into a preset intelligent analysis module, wherein each customer feedback information includes feedback data of at least one modality.

[0038] In the embodiments of this application, first, customer feedback information from multiple customers is collected using a variety of data collection sensors or API interfaces. This feedback information is diverse and includes feedback data in at least one modality, such as voice, image, and text. Specifically, in the FinTech scenario, customers can express their inquiries about financial products through voice customer service, submit text feedback about account security concerns through online forms, or report unusual transactions by uploading images of transaction receipts.

[0039] Furthermore, the non-textual feedback data in each customer feedback message is converted into textual feedback data, resulting in customer feedback information consisting solely of textual feedback data. Specifically, if a customer voice-feeds concerns about the return stability of a newly launched financial derivative product amidst market fluctuations, or uploads transaction records in the form of images with annotated abnormal fund flows, these non-textual feedback data in voice or image format are converted into textual content using technologies such as speech recognition and image recognition. For example, key transaction information in the image is converted into "[specific time] an abnormal fund outflow of [specific amount] occurred to [specific account]," and the voice message "the nurse's nighttime rounds were too long" is converted into "[specific time], [specific location], [involved person], and [specific issue]," thereby generating a full-text customer feedback message. The converted customer feedback message is then cleaned. In financial and healthcare environments, customer feedback information may contain irrelevant symbols, repeated expressions, or incorrect spellings. Pre-set cleaning rules can be used to remove this irrelevant information and correct errors in customer feedback information, improving data accuracy and usability.

[0040] Next, the cleaned customer feedback information is standardized. In the financial sector, different customers may use different terms to describe the same financial business or issue. For example, "wealth management product" may be expressed as "wealth management plan," "investment product," etc. In healthcare scenarios, "poor doctor attitude," "non-standard nurse operation," and "examination delay" can be unified into "problematic medical service attitude," "problematic medical operation compliance," and "problematic diagnosis and treatment process efficiency." Through standardization, these different colloquial expressions are unified into standard terminology, facilitating subsequent classification and analysis.

[0041] Finally, classification features are extracted from the standardized customer feedback information. Based on the extracted classification features, a preset rule engine or a pre-trained lightweight classification model is used to preliminarily categorize the feedback information and assign initial classification labels to each customer feedback message. These classification labels can include "complaint," "claim," "consultation," "suggestion," and so on.

[0042] S20. Based on a preset intelligent analysis module, determine the emotional tendency and demand content corresponding to each piece of feedback information to be processed, and determine the feedback category and processing priority corresponding to the corresponding feedback information to be processed according to the emotional tendency and demand content, and sort the multiple pieces of feedback information to be processed according to the feedback category and processing priority corresponding to each piece of feedback information to be processed.

[0043] In order to process customer feedback efficiently and accurately to improve customer service quality and optimize financial products, such as Figure 2As shown, the system, based on a preset intelligent analysis module, executes the following steps S21 to S23 to analyze the pending feedback information and then sort the multiple pending feedback information based on the analysis structure. By sorting the multiple pending feedback information, customer service personnel can clearly understand the importance and priority of the feedback, avoiding confusion and disorder in feedback processing. In this way, when faced with a large amount of customer feedback, customer service personnel can directly start processing the high-priority feedback, improving processing efficiency and thereby increasing customer satisfaction.

[0044] S21. Based on the preset intelligent analysis module, determine the emotional tendency and demand content corresponding to each piece of feedback information to be processed.

[0045] In this step, the preset analysis module uses natural language processing technology and a preset service sentiment dictionary to identify sentiment keywords and their corresponding weights, and then determine the sentiment type. It should be noted that sentiment types are mainly divided into positive, neutral, and negative. For negative sentiment, the corresponding sentiment level is further determined by the density of sentiment words and contextual semantics. Finally, the combination of the sentiment type and the corresponding sentiment level or only the sentiment type is output as the sentiment tendency. The service sentiment dictionary includes sentiment keywords and the weights corresponding to the sentiment keywords.

[0046] At the same time, using named entity recognition technology from natural language processing, we extract entity words related to the target FinTech or healthcare scenarios from the pending feedback information, such as "hospitalization allowance" (insurance product type), "claims" (service process), "payment" (service outcome), "diagnosis certificate," and "cost list" (claims document types). Deep learning semantic analysis technology is used to identify explicit and implicit needs in the pending feedback information. Based on the extracted entity words, explicit and implicit needs, a corresponding demand logic chain is generated for each pending feedback information. For example, the feedback information may read, "I submitted the claim documents for hospitalization allowance after being discharged from the hospital on April 5th, including the diagnosis certificate and cost list. Customer service said that payment would be made 15 business days after submission. As of May 20th, I still haven't received the payment." The explicit need, "I still haven't received the payment as of May 20th," directly expresses dissatisfaction with the result. The implicit need reflects the claim that the service has not lived up to its promise despite fulfilling the user's obligations. The resulting demand logic chain is: submission of documents - promised payment time - payment not made after the deadline - customer service excuses.

[0047] Combined with the feedback knowledge graph in the knowledge base, each demand logic chain is analyzed to determine its corresponding core contradiction, and a standardized description of the core contradiction is made, thereby obtaining the demand content corresponding to each feedback information to be processed. Continuing with the above feedback information as an example, "The claim materials for hospitalization allowance were submitted. After the materials were submitted, the customer service promised to make the payment within 15 working days, but the payment was not received on May 20th." The causal relationship points to the core contradiction as "complaint about the delay in the progress of hospitalization allowance claims." Then, a standardized description of the core contradiction is made. The structured description includes the customer identification field, the demand subject, and the demand description. Specifically, it can be "Customer identification: ID12345, demand subject: hospitalization allowance claim service, demand description: the payment was not made after the materials were submitted, and the customer service did not effectively resolve it."

[0048] Through the above steps, the DeepSeek module removes redundant information from the original feedback layer by layer, combines entity recognition, semantic analysis and knowledge base matching, and finally accurately extracts the core needs, providing customer service with clear processing goals, and realizing intelligent analysis and efficient use of customer feedback.

[0049] S22. Determine the feedback category and processing priority corresponding to the feedback information to be processed based on the emotional tendency and demand content.

[0050] In the embodiment of the present application, in order to achieve efficient response of health insurance customer service, such as Figure 3 As shown, the system determines the feedback category and processing priority corresponding to each feedback information to be processed by executing the following steps S221 to S223, providing data support for the subsequent system to sort multiple feedback information to be processed according to the feedback category and processing priority.

[0051] S221. For each emotional tendency and demand content corresponding to the feedback information to be processed, the initial classification label of the feedback information to be processed is identified based on the preset analysis module, the initial classification corresponding to the feedback information to be processed is obtained, and all historical feedback information corresponding to the initial classification is determined based on the feedback knowledge graph in the knowledge base.

[0052] In this step, for the emotional tendency and demand content of each feedback information to be processed, the preset analysis module will identify the initial classification label of the feedback information to be processed to determine its corresponding initial classification. In the actual operation process, the initial classification can be "complaint category", "claim category", "consultation category" and "suggestion category". Relevant technical personnel can also improve the classification rules according to actual needs, and add categories and classification features corresponding to the categories in the classification rules. Finally, based on the feedback knowledge graph in the knowledge base, all historical feedback information associated with the initial classification is searched. For example, if a customer feedback is about the inconvenience of operating a financial APP, the initial classification may be "complaint category", and all previous historical feedback on "complaint category" can be obtained through the feedback knowledge graph.

[0053] S222: Calculate the similarity between the demand content and each historical feedback information, determine the historical feedback information whose similarity meets a preset similarity threshold, and determine the historical feedback category corresponding to the historical feedback information, and use the historical feedback category as the feedback category corresponding to the feedback information to be processed.

[0054] In this step, the system calculates the similarity between the current demand content and each historical feedback information, and filters out the historical feedback information whose similarity meets the preset similarity threshold. During the actual operation, the historical feedback information with the highest similarity can be selected. When the similarity results are consistent, other similarity calculation methods can be used for comprehensive calculation to select the historical feedback information with the highest comprehensive calculation score. Furthermore, the historical feedback categories corresponding to these historical feedback information are determined, and used as the feedback categories corresponding to the current feedback information to be processed. For example, if the current feedback information to be processed is feedback about the inconvenient operation of a financial APP or medical APP and has a high similarity with the historical "APP usage experience" feedback, the current feedback will be classified into the "APP usage experience" category.

[0055] S223: Determine whether there is an emotion level in the emotion tendency, and set the priority of the feedback information to be processed according to the emotion type or the combination of the emotion tendency type and the emotion level.

[0056] In this step, the system first determines whether the sentiment level exists within the sentiment orientation. If the sentiment level does not exist, it indicates that the sentiment category is positive or neutral. The pending feedback information is not urgently processed and can be prioritized as low. For example, if a customer expresses satisfaction with the service attitude of a financial product, this positive feedback can be assigned a low priority and processed last. If the sentiment level exists within the sentiment orientation, it indicates that the sentiment category is negative. Negative feedback requires prompt processing. In this case, the system determines the frequency of historical feedback and, based on the frequency and sentiment level, prioritizes the pending feedback information as high, second, or average. Specifically, a combination of strongly negative and high-frequency issues can be assigned the highest priority. A combination of moderately negative and high-frequency issues, as well as a combination of strongly negative and low-frequency issues, can be assigned the second-highest priority. A combination of slightly negative and low-frequency issues, as well as a combination of moderately negative and low-frequency issues, can be assigned average priority. For example, if a customer expresses strong dissatisfaction with a high-risk issue in a financial product and has a high historical feedback frequency for this issue, this feedback can be assigned a high priority.

[0057] S23. Sort the plurality of feedback information to be processed according to the feedback category and processing priority corresponding to each piece of feedback information to be processed.

[0058] In this step, the system groups pending feedback information of the same feedback category into the same feedback category group based on the feedback category corresponding to each pending feedback information. Multiple feedback category groups are sorted from high to low priority based on the category priority rules stored in the knowledge base, which clearly define the priority level for each category. For example, "Insurance Claims" generally takes precedence over "Product Consultation," and "Medical Quality" takes precedence over "Service Consultation."

[0059] While sorting components, the system also needs to sort the pending feedback within each feedback category. Specifically, the system sorts the pending feedback within each feedback category from high to low priority. This allows the system to prioritize important and urgent customer feedback, improving processing efficiency and customer satisfaction.

[0060] S30. Based on the sorting results, determine the processing complexity corresponding to each piece of feedback information to be processed and the processing flow that matches the processing complexity one by one according to the feedback knowledge graph in the knowledge base, and process the feedback information to be processed according to the processing flow.

[0061] In the fields of financial technology and healthcare, in order to efficiently and accurately process customer feedback and improve customer service quality and financial business operation efficiency. After sorting the feedback information to be processed, Figure 4 As shown, the system determines the processing complexity of each pending feedback message based on the feedback knowledge graph in the knowledge base. This complexity assessment takes into account multiple factors, including the diversity of financial services and the complexity of customer needs. It then matches the processing complexity with a corresponding process, strictly following that process to ensure that every customer feedback is properly addressed.

[0062] S31. Extract feedback features corresponding to each piece of feedback information to be processed, perform semantic matching on the extracted feedback features with the nodes of the feedback knowledge graph, and determine the complexity level of the feedback information to be processed.

[0063] In this step, the system first obtains evaluation dimensions set by relevant technical personnel to quantify the processing complexity of the pending feedback information. Then, based on the evaluation dimensions, it extracts representative feedback features from each pending feedback information. These features include the type of financial product mentioned by the customer, the keywords in the feedback question, and the business process involved, or the type of medical service mentioned by the customer, the keywords in the feedback question, and the business process involved. Next, the extracted feedback features are semantically matched with nodes in the feedback knowledge graph. This matching method deeply analyzes the degree of correlation between the pending feedback information and the existing knowledge in the knowledge graph, and then determines the processing complexity of the pending feedback information. Specifically, if the correlation is high, the pending feedback information may be a common problem with low processing complexity, and the corresponding processing method and solution can be directly retrieved from the knowledge graph. If the correlation is low, the pending feedback information may be a new or complex problem with high processing complexity, requiring further analysis and processing. In FinTech and healthcare scenarios, determining the complexity level helps to rationally allocate processing resources and improve processing efficiency.

[0064] S32. Determine a processing flow for the feedback information to be processed according to the complexity level.

[0065] When it is determined that the complexity level of the feedback information to be processed is the first preset level, it means that the feedback information to be processed is relatively simple, with a high degree of universality and standardization. At this time, the predefined standardized reply templates and processing suggestions can be directly retrieved from the knowledge base. The standardized reply templates and processing suggestions in the financial technology scenario are based on a large amount of historical feedback data and financial business experience, and the standardized reply templates and processing suggestions in the medical and health scenario are based on a large amount of historical feedback data and medical service experience. Fill the customer's relevant information (such as customer name, account information, etc.) and specific processing suggestions into the standardized reply template to form personalized reply content. Finally, the filled-in standardized reply template is sent to the terminal that initiated the feedback information to be processed, that is, the customer terminal, to complete the feedback processing of the feedback information to be processed. This processing method is efficient and fast, and can meet the basic needs of customers in a timely manner.

[0066] If the complexity level of the pending feedback information is determined to be the second preset level, it indicates that the pending feedback information is relatively complex and may contain missing information or require further in-depth analysis. At this point, the intelligent analysis module is activated to analyze the pending feedback information. Using technologies such as natural language processing and machine learning, it identifies key missing information in the pending feedback information and generates guidance based on the analysis results to guide customers in providing the necessary information. Simultaneously, the system accurately retrieves service knowledge associated with the pending feedback information from the knowledge base. This service knowledge covers detailed information about the product or service, business processes, risk warnings, and other content. This guidance and service knowledge are sent to the customer service terminal, providing comprehensive information support for customer service personnel. Furthermore, the system can automatically create task tickets to record the processing progress and status of the pending feedback information in real time, facilitating tracking and management to ensure that the pending feedback information is properly handled.

[0067] S40. Record the processing flow and processing results corresponding to each piece of feedback information to be processed, and update the feedback knowledge graph in the knowledge base according to the processing flow and processing results corresponding to each piece of feedback information to be processed.

[0068] In the field of financial technology, in order to continuously optimize customer service quality and improve business processing efficiency, for each piece of pending feedback information, after processing is completed, its corresponding processing flow and processing results must be recorded in detail. Subsequently, the feedback knowledge graph in the knowledge base is updated based on these records to ensure that the knowledge graph can promptly reflect the latest business conditions and customer needs. Specifically, the processing flow and processing results corresponding to each piece of pending feedback information are first associated with the pending feedback information to form a complete information set. Next, the associated pending feedback information is stored in a designated storage space, which is specifically used to store historical data related to customer feedback for subsequent analysis and use.

[0069] When the amount of feedback information to be processed stored in the designated storage space reaches a preset value, it means that a sufficient amount of data has been accumulated and an effective knowledge update can be performed. At this time, the processing flow and processing results of each piece of feedback information to be processed are deeply identified, and new customer entities (such as a specific type of financial customer group or medical service customer group), problem entities (such as problems with new financial products or new medical services), processing entities (such as innovative processing strategies) or result entities (such as special processing effects) are extracted from them. Then, based on technical means such as co-occurrence analysis or rule engines, an association relationship is established between the new entity and the existing entity in the feedback knowledge graph, or the confidence of the association relationship between the existing entities is weighted and updated. In this way, the feedback knowledge graph is continuously enriched and improved to better serve the decision-making and processing of financial business.

[0070] In addition, the system can also locate inefficient links by processing statistical indicators of data (such as the average processing time of each stage, customer satisfaction distribution, and problem repetition rate). For example: If the "supervisor review" stage takes an average of 48 hours and the problem-solving rate at this stage is low, it is identified as a redundant link. Delete the redundant link in the relevant processing flow. And identify high-value processing actions by querying the relationship weights of "processing entity-result entity" in the knowledge graph. For example, the customer satisfaction of "transfer to expert customer service" is 92%, while "calling template reply" is only 65%. It can be recommended to reduce the applicable scenarios of template replies, thereby achieving autonomous optimization of the corresponding service processes.

[0071] The method provided in the embodiments of the present application determines the emotional tendency and demand content of each pending feedback message through a preset intelligent analysis module, enabling a more in-depth and accurate understanding of the essence of customer feedback, avoiding the subjective bias and misunderstanding that may occur in manual analysis. Furthermore, by sorting multiple pending feedback messages according to feedback category and processing priority, the embodiments of the present application enable processing personnel to clearly understand the importance and priority of the feedback, avoiding confusion and disorder in feedback processing. In this way, when faced with a large amount of customer feedback, processing personnel can directly start processing the high-priority feedback, improving processing efficiency. Moreover, for pending feedback messages of different complexities, the present application can adopt different processing methods. For simple feedback, an automated process can be used for rapid processing, while for complex feedback, professional personnel can be arranged for in-depth analysis and processing, improving the pertinence and effectiveness of the processing. In addition, the knowledge base provides customer service personnel with rich knowledge support, improving service efficiency. At the same time, the self-learning capability enables the system to continuously update the knowledge base based on new customer feedback and processing experience, keeping pace with the times and improving service quality and accuracy.

[0072] Further, as Figure 1 In the specific implementation of the method, the embodiment of the present application provides a feedback information processing device, such as Figure 5 As shown, the system includes: a collection module 501, an analysis module 502, a processing module 503, and an optimization module 504.

[0073] The collection module 501 is configured to collect multiple customer feedback information, perform data preprocessing on the multiple customer feedback information to obtain multiple pieces of feedback information to be processed, and input the multiple pieces of feedback information to be processed into a preset intelligent analysis module, wherein each piece of customer feedback information includes feedback data of at least one modality;

[0074] The analysis module 502 is configured to determine, based on the preset intelligent analysis module, the emotional tendency and demand content corresponding to each piece of feedback information to be processed, determine the feedback category and processing priority corresponding to the corresponding piece of feedback information to be processed according to the emotional tendency and demand content, and sort the plurality of pieces of feedback information to be processed according to the feedback category and processing priority corresponding to each piece of feedback information to be processed;

[0075] The processing module 503 is configured to determine, based on the sorting result and according to the feedback knowledge graph in the knowledge base, the processing complexity corresponding to each piece of feedback information to be processed, and a processing flow matching the processing complexity, and process the feedback information to be processed according to the processing flow;

[0076] The optimization module 504 is used to record the processing flow and processing results corresponding to each piece of feedback information to be processed, and update the feedback knowledge graph in the knowledge base according to the processing flow and processing results corresponding to each piece of feedback information to be processed.

[0077] In a specific application scenario, the collection module 501 is used to collect multiple customer feedback information through multiple data collection sensors or API interfaces; convert the non-text modality feedback data in each customer feedback information into text modality feedback data to obtain customer feedback information that only includes text modality feedback data; clean the converted customer feedback information to remove irrelevant information in the customer feedback information and modify erroneous content in the customer feedback information; standardize the cleaned customer feedback information and extract classification features from the standardized customer feedback information; perform preliminary classification based on the extracted classification features based on a preset rule engine or a pre-trained lightweight classification model, and set an initial classification label for the customer feedback information.

[0078] In a specific application scenario, the analysis module 502 is used to identify the emotion type in each of the feedback information to be processed based on the preset analysis module, through natural language processing technology and a preset service emotion dictionary, and for negative emotions in the emotion type, judge the emotion level corresponding to the negative emotion through emotion word density and context semantics, and output the combination of the emotion type and the corresponding emotion level or the emotion type as the emotion tendency, wherein the emotion type includes positive emotion, neutral emotion and negative emotion, and the service emotion dictionary includes emotion keywords and weights corresponding to emotion keywords; extract entity words related to the target scenario from each of the feedback information to be processed through named entity recognition technology of natural language processing, and use deep learning semantic analysis technology to identify explicit needs and implicit needs in each of the feedback information to be processed, and generate a corresponding demand logic chain for each of the feedback information to be processed based on the entity words, the explicit needs and the implicit needs; combine the feedback knowledge graph in the knowledge base to determine the core contradiction corresponding to each demand logic chain, perform a standardized description of the core contradiction, and obtain the demand content corresponding to each of the feedback information to be processed.

[0079] In a specific application scenario, the analysis module 502 is used to identify the initial classification label of the feedback information to be processed based on the preset analysis module for the emotional tendency and demand content corresponding to each of the feedback information to be processed, obtain the initial classification corresponding to the feedback information to be processed, and determine all historical feedback information corresponding to the initial classification based on the feedback knowledge graph in the knowledge base; calculate the similarity between the demand content and each historical feedback information, determine the historical feedback information whose similarity meets the preset similarity threshold, and determine the historical feedback category corresponding to the historical feedback information, and use the historical feedback category as the feedback category corresponding to the feedback information to be processed; determine whether there is an emotional level in the emotional tendency; if the emotional level does not exist in the emotional tendency, the emotional category corresponding to the emotional tendency is positive emotion or neutral emotion, and the priority of the feedback information to be processed is set to low priority; if the emotional level exists in the emotional tendency, determine the feedback frequency of the historical feedback information, and set the priority of the feedback information to be processed to high priority, second highest priority or general priority according to the feedback frequency and the emotional level.

[0080] In a specific application scenario, the analysis module 502 is used to divide the feedback information to be processed of the same feedback category into the same feedback category group according to the feedback category corresponding to each piece of feedback information to be processed; sort the multiple feedback category groups from high to low priority according to the category priority rule stored in the knowledge base, where the category priority rule is used to indicate the priority corresponding to each category; and sort the feedback information to be processed within each feedback category group from high to low according to the processing priority.

[0081] In a specific application scenario, the processing module 503 is used to extract feedback features corresponding to each of the feedback information to be processed, perform semantic matching on the extracted feedback features with nodes of the feedback knowledge graph, and determine the complexity level of the feedback information to be processed; when it is determined that the complexity level of the feedback information to be processed is a first preset level, retrieve a predefined standardized response template and processing suggestions from the knowledge base, fill the customer information corresponding to the feedback information to be processed and the processing suggestions into the standardized response template, and send the filled-in standardized response template to the client terminal to complete the feedback processing of the feedback information to be processed, where the client terminal is the terminal that initiated the customer feedback information; when it is determined that the complexity level of the feedback information to be processed is a second preset level, use the intelligent analysis module to analyze the feedback information to be processed, identify missing information in the feedback information to be processed and generate corresponding guidance content, obtain service knowledge associated with the feedback information to be processed from the knowledge base, send the guidance content and the service knowledge to the customer service terminal, create a task ticket, and update the processing status of the feedback information to be processed in real time.

[0082] In a specific application scenario, the optimization module 504 is used to associate the processing flow and processing results corresponding to each of the feedback information to be processed with the corresponding feedback information to be processed, and store the associated feedback information to be processed in a designated storage space; when the number of feedback information to be processed stored in the designated storage space meets a preset value, identify the processing flow and processing results of each of the feedback information to be processed, extract new customer entities, problem entities, processing entities or result entities from the processing flow and the processing results, and establish an association relationship between the new entity and the existing entity in the feedback knowledge graph based on co-occurrence analysis or a rule engine, or perform a weighted update on the confidence of the association relationship between existing entities.

[0083] The device provided by the embodiment of the present application determines the emotional tendency and demand content of each pending feedback message through a preset intelligent analysis module, enabling a more in-depth and accurate understanding of the essence of customer feedback, avoiding the subjective bias and misunderstanding that may occur in manual analysis. Furthermore, by sorting multiple pending feedback messages according to feedback category and processing priority, the embodiment of the present application allows processing personnel to clearly understand the importance and priority of the feedback, avoiding confusion and disorder in feedback processing. In this way, when faced with a large amount of customer feedback, processing personnel can directly start processing from the high-priority feedback, improving processing efficiency. Moreover, for pending feedback messages of different complexities, the present application can adopt different processing methods. For simple feedback, an automated process can be used for rapid processing, while for complex feedback, professional personnel can be arranged for in-depth analysis and processing, improving the pertinence and effectiveness of the processing. In addition, the knowledge base provides customer service personnel with rich knowledge support, improving service efficiency. At the same time, the self-learning capability enables the system to continuously update the knowledge base based on new customer feedback and processing experience, keeping pace with the times and improving service quality and accuracy.

[0084] It should be noted that for other corresponding descriptions of the functional units involved in the feedback information processing device provided in the embodiment of the present application, please refer to Figures 1 to 4 The corresponding description in will not be repeated here.

[0085] To solve the above technical problems, the embodiment of the present invention also provides a computer device. Figure 6 , Figure 6 This is a basic structural block diagram of the computer device in this embodiment.

[0086] like Figure 6 As shown, a schematic diagram of the internal structure of a computer device. The computer device includes a processor, a non-volatile storage medium, a memory and a network interface connected via a system bus. Among them, the non-volatile storage medium of the computer device stores an operating system, a database and computer-readable instructions, and the database may store a control information sequence. When the computer-readable instructions are executed by the processor, the processor can implement a data relationship reconstruction method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor can execute a data relationship reconstruction method. The network interface of the computer device is used to connect and communicate with the terminal. Those skilled in the art will understand that Figure 6 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0087] In this embodiment, the processor is used to execute Figure 5 The memory stores the program codes and various data required to execute the above modules. The network interface is used to transmit data between user terminals or servers.

[0088] The present invention also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the data relationship reconstruction method in any of the above embodiments.

[0089] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed, the program can include the processes in the above-described method embodiments. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0090] The present invention also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the data relationship reconstruction method in any of the above embodiments.

[0091] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed, the program can include the processes in the above-described method embodiments. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0092] Those skilled in the art will appreciate that the steps, measures, and schemes in the various operations, methods, and processes discussed in this application may be interchanged, modified, combined, or deleted. Furthermore, other steps, measures, and schemes in the various operations, methods, and processes discussed in this application may also be interchanged, modified, rearranged, decomposed, combined, or deleted. Furthermore, steps, measures, and schemes in the prior art that are similar to those disclosed in this application may also be interchanged, modified, rearranged, decomposed, combined, or deleted.

[0093] The above description is only part of the implementation methods of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A feedback information processing method, characterized in that: include: Collecting multiple customer feedback information, performing data preprocessing on the multiple customer feedback information to obtain multiple feedback information to be processed, and inputting the multiple feedback information to be processed into a preset intelligent analysis module, wherein each of the customer feedback information includes feedback data of at least one modality; Determining, based on the preset intelligent analysis module, the emotional tendency and demand content corresponding to each of the feedback information to be processed, and determining the feedback category and processing priority corresponding to the corresponding feedback information to be processed according to the emotional tendency and demand content, and sorting the plurality of feedback information to be processed according to the feedback category and processing priority corresponding to each of the feedback information to be processed; Based on the ranking results, determine the processing complexity corresponding to each piece of feedback information to be processed and the processing flow that matches the processing complexity according to the feedback knowledge graph in the knowledge base, and process the feedback information to be processed according to the processing flow; Record the processing flow and processing results corresponding to each of the feedback information to be processed, and update the feedback knowledge graph in the knowledge base according to the processing flow and processing results corresponding to each of the feedback information to be processed.

2. The method according to claim 1, characterized in that The collecting of multiple customer feedback information and performing data preprocessing on the multiple customer feedback information includes: Collecting multiple customer feedback information through multiple data collection sensors or API interfaces; Converting non-text modal feedback data in each customer feedback information into text modal feedback data to obtain customer feedback information including only text modal feedback data, cleaning the converted customer feedback information to remove irrelevant information in the customer feedback information, and correcting erroneous content in the customer feedback information; The cleaned customer feedback information is standardized, and classification features are extracted from the standardized customer feedback information. Based on a preset rule engine or a pre-trained lightweight classification model, preliminary classification is performed according to the extracted classification features, and an initial classification label is set for the customer feedback information.

3. The method according to claim 1, characterized in that The determining of the emotional tendency and demand content corresponding to each piece of feedback information to be processed includes: Based on the preset analysis module, the emotional type in each of the feedback information to be processed is identified through natural language processing technology and a preset service emotional dictionary, and for negative emotions in the emotional type, the emotional level corresponding to the negative emotion is determined through emotional word density and contextual semantics, and a combination of the emotional type and the corresponding emotional level or the emotional type is output as the emotional tendency, wherein the emotional type includes positive emotion, neutral emotion and the negative emotion, and the service emotional dictionary includes emotional keywords and weights corresponding to the emotional keywords; Extracting entity words related to the target scenario from each piece of feedback information to be processed by using named entity recognition technology based on natural language processing, identifying explicit and implicit requirements in each piece of feedback information to be processed by using deep learning semantic analysis technology, and generating a corresponding demand logic chain for each piece of feedback information to be processed based on the entity words, the explicit requirements, and the implicit requirements; Combined with the feedback knowledge graph in the knowledge base, the core contradiction corresponding to each demand logic chain is determined, the core contradiction is described in a standardized manner, and the demand content corresponding to each of the feedback information to be processed is obtained.

4. The method according to claim 3, characterized in that Determining the feedback category and processing priority corresponding to the feedback information to be processed based on the emotional tendency and demand content includes: For each of the sentiment tendencies and demand contents corresponding to the feedback information to be processed, the preset analysis module identifies the initial classification label of the feedback information to be processed, obtains the initial classification corresponding to the feedback information to be processed, and determines all historical feedback information corresponding to the initial classification based on the feedback knowledge graph in the knowledge base; Calculating the similarity between the demand content and each historical feedback information, determining the historical feedback information whose similarity meets a preset similarity threshold, and determining the historical feedback category corresponding to the historical feedback information, and using the historical feedback category as the feedback category corresponding to the feedback information to be processed; determining whether there is an emotional level in the emotional tendency; If the emotional level does not exist in the emotional tendency, the emotional category corresponding to the emotional tendency is positive emotion or neutral emotion, and the priority of the feedback information to be processed is set to low priority; If the emotional level exists in the emotional tendency, the feedback frequency of the historical feedback information is determined, and the priority of the feedback information to be processed is set to high priority, second highest priority or general priority according to the feedback frequency and the emotional level.

5. The method according to claim 4, characterized in that The sorting of the plurality of feedback information to be processed according to the feedback category and processing priority corresponding to each piece of feedback information to be processed includes: According to the feedback category corresponding to each piece of feedback information to be processed, the feedback information to be processed of the same feedback category is divided into the same feedback category group; sorting the plurality of feedback category groups from high to low priority according to a category priority rule stored in the knowledge base, wherein the category priority rule is used to indicate the priority corresponding to each category; The pending feedback information in each feedback category group is sorted from high to low according to the processing priority.

6. The method according to claim 1, characterized in that The step of determining the processing complexity corresponding to each piece of feedback information to be processed and a processing flow matching the processing complexity according to the feedback knowledge graph in the knowledge base, and processing the feedback information to be processed according to the processing flow, includes: Extracting feedback features corresponding to each of the feedback information to be processed, performing semantic matching on the extracted feedback features with nodes of the feedback knowledge graph, and determining the complexity level of the feedback information to be processed; When it is determined that the complexity level of the feedback information to be processed is the first preset level, a predefined standardized response template and processing suggestion are retrieved from the knowledge base, the customer information corresponding to the feedback information to be processed and the processing suggestion are filled into the standardized response template, and the filled standardized response template is sent to the client terminal to complete the feedback processing of the feedback information to be processed, where the client terminal is the terminal that initiated the customer feedback information; When it is determined that the complexity level of the feedback information to be processed is the second preset level, the intelligent analysis module is used to analyze the feedback information to be processed, identify missing information in the feedback information to be processed and generate corresponding guidance content, and obtain service knowledge associated with the feedback information to be processed in the knowledge base, send the guidance content and the service knowledge to the customer service terminal, create a task work order and update the processing status of the feedback information to be processed in real time.

7. The method according to claim 5, characterized in that The updating of the feedback knowledge graph in the knowledge base according to the processing flow and processing result corresponding to each of the feedback information to be processed includes: Associating the processing flow and processing result corresponding to each piece of feedback information to be processed with the corresponding feedback information to be processed, and storing the associated feedback information to be processed in a designated storage space; When the number of unprocessed feedback information stored in the designated storage space meets a preset value, the processing flow and processing results of each of the unprocessed feedback information are identified, new customer entities, problem entities, processing entities or result entities are extracted from the processing flow and the processing results, and based on co-occurrence analysis or a rule engine, an association relationship between the new entity and the existing entity in the feedback knowledge graph is established, or the confidence of the association relationship between the existing entities is weighted and updated.

8. A feedback information processing device, characterized in that: include: a collection module configured to collect a plurality of customer feedback information, perform data preprocessing on the plurality of customer feedback information to obtain a plurality of feedback information to be processed, and input the plurality of feedback information to be processed into a preset intelligent analysis module, wherein each of the customer feedback information includes feedback data of at least one modality; an analysis module configured to determine, based on the preset intelligent analysis module, the emotional tendency and demand content corresponding to each piece of feedback information to be processed, determine the feedback category and processing priority corresponding to the corresponding piece of feedback information to be processed according to the emotional tendency and demand content, and sort the plurality of pieces of feedback information to be processed according to the feedback category and processing priority corresponding to each piece of feedback information to be processed; A processing module is configured to determine, based on the sorting result and according to the feedback knowledge graph in the knowledge base, the processing complexity corresponding to each piece of feedback information to be processed, and a processing flow matching the processing complexity, and process the feedback information to be processed according to the processing flow; The optimization module is used to record the processing flow and processing results corresponding to each of the feedback information to be processed, and update the feedback knowledge graph in the knowledge base according to the processing flow and processing results corresponding to each of the feedback information to be processed.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images