Method and system for intelligently collecting and responding to customer feedback

Abstract

The invention discloses a method and system for intelligently collecting and responding to customer feedback, and the method comprises the steps: S1, obtaining original complaint data from at least one Internet platform through a preset Internet data interface; s2, performing information extraction and structured processing on the original complaint data to obtain complaint feature information which can be queried; s3, querying an internal business system based on the complaint feature information to obtain matched and associated order data and user attribute information, and generating an initial processing scheme for the complaint according to a preset processing strategy database; s4, generating an adjustable interaction strategy, communicating with the user through voice call-out, and dynamically adjusting the interaction process based on the real-time feedback of the user during communication; and S5, executing subsequent operation based on a communication result. The method has the effects of process automation, intelligent decision making, emotion perception, dynamic interaction, effective and timely public opinion restraint and the like.

Description

Technical Field

[0001] This invention relates to the field of customer service technology, and in particular to a method and system for intelligently collecting and responding to customer feedback. Background Technology

[0002] With the widespread use of the internet, users are increasingly accustomed to filing complaints on various online platforms to protect their rights. Currently, businesses primarily rely on manual methods to handle such complaints, including: manual monitoring: requiring customer service personnel to continuously monitor various public opinion platforms, which is inefficient and prone to omissions; manual verification: after a complaint is discovered, information such as order details and user details must be manually extracted from unstructured text / images / videos and verified against internal systems, a tedious and time-consuming process; and manual decision-making and communication: customer service needs to manually formulate compensation plans based on factors such as the user's historical value and order details, and communicate with the user by phone, which is costly and slow. This highly manual approach leads to a severely untimely response to complaints. Summary of the Invention

[0003] To address the technical problems existing in the background art, this invention proposes a method and system for intelligently collecting and responding to customer feedback. It integrates public opinion monitoring, intelligent information extraction, intelligent decision-making, and automatic outbound calling capabilities to achieve full-process automation from complaint discovery to resolution. This greatly improves the efficiency and timeliness of complaint handling, quickly soothes user emotions, curbs the impact of negative public opinion in its infancy, and reduces the human resource costs for enterprises in complaint handling.

[0004] In a first aspect, the present invention proposes a method and system for intelligently collecting and responding to customer feedback, the method comprising: S1. Obtain raw complaint data from at least one internet platform through a pre-defined internet data interface; S2. Extract and structure the information from the original complaint data to obtain queryable complaint feature information; S3. Based on the complaint feature information, query the internal business system to obtain matching and related order data and user attribute information, and generate an initial processing plan for the complaint according to the preset processing strategy database. S4. Combining the complaint feature information, the user attribute information, and the initial processing plan, generate an adjustable interaction strategy, communicate with the user via voice outbound calls, and dynamically adjust the interaction process based on the user's real-time feedback during the communication. S5. Execute subsequent operations based on communication results: If the user accepts the initial handling plan, the complaint status will be automatically updated and the initial handling plan will be executed; otherwise, a task to be manually followed up will be automatically generated and associated with the information of this handling process.

[0005] Preferably, in S2, "extracting and structuring the original complaint data" includes: Extracting entity information from text using natural language processing techniques; and / or extracting text information from images using optical character recognition techniques; and / or extracting text information from audio or video using speech recognition techniques.

[0006] Preferably, in S4, "generating an adjustable interaction strategy" includes: A pre-defined dialogue strategy library is used to construct dialogue scripts containing the main logic and multiple possible branch dialogue paths.

[0007] Preferably, when constructing a dialogue script, the corresponding communication script template is matched from the dialogue strategy library in combination with the user attribute information; the user attribute information includes user level and / or user's historical complaint frequency.

[0008] Preferably, the dialogue script includes a reassurance branch path triggered when the user rejects the initial handling plan, and the reassurance branch path includes preset emotional reassurance language, escalation commitments, and estimated subsequent handling time.

[0009] Preferably, in S4, "communicating with users via voice outbound calls" includes: It uses speech synthesis technology to communicate with users and interprets user intent through real-time speech recognition and semantic understanding technology.

[0010] Preferably, in S4, the "dynamically adjusting the interaction process based on real-time user feedback during communication" includes: Real-time identification of the emotional state and semantic content of user feedback; When the intensity of negative emotions exceeds a preset threshold, a preset emotion-soothing strategy is triggered.

[0011] Preferably, in S5, "automatically generating tasks to be manually followed up and associating them with the current processing information" includes: Create a high-priority customer work order and assign it to a human customer service specialist. The associated processing information includes: the complaint feature information, the initial processing plan and its generation basis, and the complete recording of this voice communication.

[0012] Secondly, the present invention proposes a system for intelligently collecting and responding to customer feedback, comprising: The public opinion data collection module is configured to perform the steps described in S1 of the first aspect; The information intelligent extraction module is configured to perform the steps described in S2 of the first aspect; The intelligent decision-making module is configured to perform the steps described in S3 of the first aspect; A dynamic interaction module is configured to perform the steps described in S4 of the first aspect; The task scheduling module is configured to perform the steps described in S5 of the first aspect.

[0013] This invention has the following significant advantages: 1. Full-process automation: It realizes full-link automation from "discovering complaints" to "proactive outbound call processing", freeing people from tedious work of monitoring screens, verifying and communicating, and shortening the response time from "hours" to "minutes".

[0014] 2. Intelligent Decision Making: By integrating data from complaints, business operations, and users, compensation decisions become more objective, accurate, and personalized, protecting user rights while avoiding excessive consumption of corporate resources.

[0015] 3. Emotional perception and interaction capabilities: During outbound calls, the interaction process can be dynamically adjusted based on the user's real-time feedback, making the interaction experience closer to that of a real person and greatly improving the success rate of complaint settlement.

[0016] 4. Effectively curb public opinion: By proactively and efficiently intervening in the first instance, it is possible to greatly quell user grievances and prompt users to withdraw their lawsuits, thereby minimizing potential negative public opinion risks and protecting the company's brand reputation. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for intelligently collecting and responding to customer feedback proposed in this invention. Figure 2 This is a module configuration diagram of the intelligent system for collecting and responding to customer feedback proposed in this invention. Detailed Implementation

[0018] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings and embodiments. The embodiments are implemented based on the technical solution of the present invention, providing detailed implementation methods and specific operating procedures. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them, and the scope of protection of the present invention is not limited to the following embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0019] Firstly, referring to Figure 1 This invention proposes an intelligent complaint handling method, which is executed by a system that intelligently collects and responds to customer feedback (referred to as an "intelligent agent"). The method includes: S1. Obtain raw complaint data from at least one internet platform through a pre-defined internet data interface. The internet platform can be a dedicated complaint platform such as Black Cat Complaints, or a self-media platform such as Xiaohongshu, Toutiao, or Weibo. Through continuous monitoring, once the system identifies complaint content related to the target entity (such as this company), it collects at least one type of unstructured data from the relevant text, images, and videos.

[0020] It should be noted that this invention does not limit the specific technical means of data collection; its core lies in its ability to provide original complaint data from multiple platforms in real time.

[0021] In some embodiments, a self-built distributed crawler system can be used, such as using crawler tools to crawl web pages, scanning target pages at a fixed frequency, and identifying duplicate complaints based on content hashes or unique IDs by setting keywords such as company names, product names, and brand names. In other embodiments, the function of collecting public opinion data can also be achieved by calling commercial public opinion monitoring services (such as Alibaba Cloud Public Opinion Analysis API).

[0022] S2. Extract and structure the information from the original complaint data to obtain queryable complaint feature information.

[0023] Specifically, after receiving complaint data, for text content, semantic information is extracted using natural language processing (NLP); for image content, information is extracted using optical character recognition (CV); and for audio or video content, speech content is converted into text information using speech recognition technology.

[0024] By combining one or more of the above methods, the intelligent agent transforms unstructured user descriptions into a structured set of information containing clearly defined fields, which becomes queryable complaint feature information.

[0025] For example, when a user posts a complaint on Weibo with screenshots, the intelligent agent extracts the order number from the text and the flight information from the image, thus determining that the complaint is a "flight delay complaint".

[0026] For example, when a user posts a video complaint on Xiaohongshu, the intelligent agent extracts keywords from the audio-to-text conversion and identifies the scene of the conflict from the video frames, thus determining that the complaint is a "service attitude complaint".

[0027] For example, if a user posts a text message on the Black Cat Complaint Platform stating, "I booked three tickets for a flight from Beijing to Shanghai on December 1st, paying 1800 yuan, but the platform quietly refunded the money without issuing the tickets," the Agent will extract the following structured information: Date: 2025-12-01 Trip Locations: Beijing - Shanghai Number of passengers: 3 Order amount: 1800 Order status: Refund received Complaint scenario: Ticketing failure / unjustified refund This step transforms unstructured user descriptions into queryable structured fields to facilitate information retrieval and decision-making in subsequent steps.

[0028] S3. Based on the complaint feature information, query the internal business system to obtain matching and related order data and user attribute information. Order data includes the orderer's name and contact information, while user attribute information includes one or more of the following: user level, historical spending amount, and historical complaint frequency. Subsequently, the intelligent agent performs logical calculations based on a preset processing strategy database, combining the complaint scenario, order information, and user value to generate a personalized initial processing plan for the complaint. For example, it might generate a plan to compensate with a 200 yuan no-threshold coupon.

[0029] Specifically, the compensation amount = S × V × P × F × C; in: S represents the scenario coefficient, with customer service issues taking a value of 1.0 and functional issues taking a value of 1.2.

[0030] V represents the user value coefficient, which is used to reflect the differentiated service weights for users of different levels; V = V_base + V_loyalty V_base represents the base value coefficient: the coefficient varies for different membership levels and can be set according to needs; V_loyalty = n × min(5, number of consecutive years of consumption), represents the loyalty bonus, where n is a constant and the upper limit of "number of consecutive years of consumption" is 5 years.

[0031] P represents the order amount coefficient, used to adjust the order value proportionally. F represents the user complaint frequency coefficient, which is used to take into account the moderating effect of users' historical complaint behavior;

[0032] The number of complaints in the last 30 days, weighted. It is a constant; Total number of complaints in history, weighted It is a constant; F_max is the maximum penalty coefficient, and its value is a constant.

[0033] C represents the basic compensation unit, which is a preset constant.

[0034] S4. Combining the complaint feature information, the user attribute information, and the initial processing plan, generate an adjustable interaction strategy, and communicate with the user via voice outbound calls, dynamically adjusting the interaction process based on the user's real-time feedback during the communication.

[0035] Specifically, generating adjustable interaction strategies includes: Intelligent Agent Preset Dialogue Strategy Library; Based on current complaint characteristics, user attribute information, and initial handling plan, a dialogue script containing main logic and multiple possible branch dialogue paths is constructed from the dialogue strategy library.

[0036] The dialogue script includes a reassurance branch path triggered when the user rejects the initial handling plan. The reassurance branch path includes preset emotional reassurance phrases, escalation promises, and estimated subsequent handling time.

[0037] During voice interaction, the intelligent agent uses speech synthesis technology to communicate with the user and supports user interruptions. It analyzes the user's intent through real-time speech recognition and semantic understanding. Simultaneously, the intelligent agent identifies the user's emotional state and semantic content in real time. When the intensity of negative emotions exceeds a preset threshold, it triggers a pre-defined emotion-soothing strategy, dynamically adjusting the subsequent dialogue path.

[0038] S5. Based on the communication results, automatically determine whether the user has "accepted the solution" and "agreed to withdraw the lawsuit," and then proceed with subsequent actions: If the user accepts the initial handling solution, the complaint status will be automatically updated, marked as "resolved", and the initial handling solution will be executed, such as calling the coupon system to issue compensation, thus completing the entire process. If the user does not accept the initial handling plan, a task awaiting human follow-up is automatically generated, and the current handling process information is automatically associated with that task. Specifically, a high-priority customer work order is created and assigned to a human customer service specialist. Simultaneously, the intelligent agent associates the customer work order with the current handling process information, including: the complaint characteristics, the initial handling plan and its generation basis, and the complete recording of the current voice communication, to enable efficient handling of the complaint by human customer service.

[0039] In one specific embodiment, the dialogue strategy includes: (1) Opening and apology: state your identity, mention the source of the complaint, and express your apology and concern.

[0040] (2) Propose an initial solution: clearly state the compensation plan.

[0041] (3) Proactive response: If the user agrees, guide the user to withdraw the complaint.

[0042] (4) Handling negative responses: If the user disagrees or becomes emotionally agitated, initiate an emotional reassurance strategy, promise to escalate the handling and provide a specific timeframe for follow-up contact to avoid disappointing the user.

[0043] Secondly, such as Figure 2 As shown, a system (intelligent agent) for intelligently collecting and responding to customer feedback is used to implement the method in the first aspect described above. The system includes: The public opinion data collection module is configured to execute the steps described in S1. Its main function is to collect original complaint data related to the company from one or more internet platforms.

[0044] The intelligent information extraction module is configured to execute the steps described in S2. It is primarily used to receive raw complaint data and output queryable complaint feature information, enabling the identification and formatting of key information from unstructured text, images, and audio.

[0045] The intelligent decision-making module is configured to execute the steps described in S3. Its main function is to receive complaint characteristic information, obtain relevant data by accessing the internal business database, perform logical operations based on a preset processing strategy database, and finally output a targeted initial processing plan.

[0046] The dynamic interaction module is configured to execute the steps described in S4. It is primarily used to combine complaint feature information, user attribute information, and the initial processing plan to generate an interaction strategy, and then engage in dynamically adjustable voice dialogue with the user through integrated speech synthesis, semantic understanding, and emotion recognition units.

[0047] The task scheduling module is configured to execute the steps described in S5. Its main function is to automatically perform status updates or task creation and distribution based on the results of communication and interaction.

[0048] In summary, the present invention has the following significant advantages: 1. Full-process automation: It realizes full-link automation from "discovering complaints" to "proactive outbound call processing", freeing people from tedious work of monitoring screens, verifying and communicating, and shortening the response time from "hours" to "minutes".

[0049] 2. Intelligent Decision Making: By integrating multi-source data from complaints, business operations, and users and applying intelligent decision-making, compensation decisions become more objective, accurate, and personalized, protecting user rights while avoiding excessive consumption of corporate resources.

[0050] 3. Emotional perception and interaction capabilities: During outbound calls, the interaction process can be dynamically adjusted based on the user's real-time feedback, making the interaction experience closer to that of a real person and greatly improving the success rate of complaint settlement.

[0051] 4. Effectively curb public opinion: By proactively and efficiently intervening in the first instance, it is possible to greatly quell user grievances and prompt users to withdraw their lawsuits, thereby minimizing potential negative public opinion risks and protecting the company's brand reputation.

[0052] As used in this invention, "an embodiment" or "embodiment" refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the invention. The terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such processes, methods, products, or apparatus.

[0053] This specification provides method operation steps as shown in the embodiments or flowcharts, but based on conventional or non-inventive labor, more or fewer operation steps may be included. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only execution order. In actual system or server products, the method can be executed in the order shown in the embodiments or drawings, or in parallel (e.g., in a parallel processor or multi-threaded processing environment), or the execution order of steps without timing constraints can be adjusted.

[0054] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for intelligently collecting and responding to customer feedback, characterized in that the method... include: S1. Obtain raw complaint data from at least one internet platform through a pre-defined internet data interface; S2. Extract and structure the information from the original complaint data to obtain queryable complaint feature information; S3. Based on the complaint feature information, query the internal business system to obtain matching and related order data and user attribute information, and generate an initial processing plan for the complaint according to the preset processing strategy database. S4. Combining the complaint feature information, the user attribute information, and the initial processing plan, generate an adjustable interaction strategy, communicate with the user via voice outbound calls, and dynamically adjust the interaction process based on the user's real-time feedback during the communication. S5. Execute subsequent operations based on communication results: If the user accepts the initial handling plan, the complaint status will be automatically updated and the initial handling plan will be executed; otherwise, a task to be manually followed up will be automatically generated and associated with the information of this handling process.

2. The method for intelligently collecting and responding to customer feedback according to claim 1, characterized in that, In S2, "extracting and structuring the original complaint data" includes: Extracting entity information from text using natural language processing techniques; and / or extracting text information from images using optical character recognition techniques; and / or extracting text information from audio or video using speech recognition techniques.

3. The method for intelligently collecting and responding to customer feedback according to claim 1, characterized in that, In S4, "generating adjustable interaction strategies" includes: A pre-defined dialogue strategy library is used to construct dialogue scripts containing the main logic and multiple possible branch dialogue paths.

4. The method for intelligently collecting and responding to customer feedback according to claim 3, characterized in that, When constructing a dialogue script, the corresponding communication script template is matched from the dialogue strategy library based on the user attribute information; the user attribute information includes user level and / or user's historical complaint frequency.

5. The method for intelligently collecting and responding to customer feedback according to claim 3 or 4, characterized in that, The dialogue script includes a reassurance branch path triggered when the user rejects the initial handling plan. The reassurance branch path includes preset emotional reassurance phrases, escalation promises, and estimated subsequent handling time.

6. The method for intelligently collecting and responding to customer feedback according to claim 1, characterized in that, In S4, "communicating with users via voice outbound calls" includes: It uses speech synthesis technology to communicate with users and interprets user intent through real-time speech recognition and semantic understanding technology.

7. The method for intelligently collecting and responding to customer feedback according to claim 1, characterized in that, In S4, the "dynamically adjusting the interaction process based on real-time user feedback during communication" includes: Real-time identification of the emotional state and semantic content of user feedback; When the intensity of negative emotions exceeds a preset threshold, a preset emotion-soothing strategy is triggered.

8. The method for intelligently collecting and responding to customer feedback according to claim 1, characterized in that, In S5, "automatically generating tasks requiring manual follow-up and associating them with current processing information" includes: Create a high-priority customer work order and assign it to a human customer service specialist. The associated processing information includes: the complaint feature information, the initial processing plan and its generation basis, and the complete recording of this voice communication.

9. A system for intelligently collecting and responding to customer feedback, characterized in that, include: The public opinion data collection module is configured to perform the steps described in S1 of claim 1. The information intelligent extraction module is configured to perform the steps described in S2 of claim 1; The intelligent decision-making module is configured to perform the steps described in S3 of claim 1; A dynamic interaction module, configured to perform the steps described in S4 of claim 1; The task scheduling module is configured to perform the steps described in S5 of claim 1.

Images