Intelligent customer service method and system

Abstract

An intelligent customer service method includes the following steps: receiving demand information and auxiliary information generated by a user in an interactive module, and then generating demand type data and user scenario data; using vector indexing technology to match a first text group in a database module based on these two types of data; generating response text using a natural language processing model based on the demand information and the first text group; and providing a reply to the user through the interactive module based on the response text.

Description

Technical Field

[0001] This invention relates to the field of intelligent customer service technology, and in particular to an intelligent customer service method and system. Prior Technology

[0002] In today's digital age, with the rapid development of artificial intelligence (AI) and natural language processing (NLP) technologies, intelligent customer service systems are becoming increasingly prevalent across various industries. Traditional customer service systems primarily rely on human customer service representatives to answer user questions. However, human customer service suffers from high costs, low efficiency, and inconsistent service quality, and cannot effectively handle the large volume of questions from users. The emergence of intelligent customer service systems fills this gap. By combining natural language processing, machine learning, and database technologies, they achieve automatic identification and intelligent responses to user questions, significantly improving the efficiency and quality of user service.

[0003] Some customer service systems in Xizhi Technology primarily use rule-based or template-based algorithms to generate response text. Rule-based algorithms require manually defining a large number of rules to meet various user needs, but the maintenance and updating costs of these rules are high. Template-based algorithms can quickly generate response text, but the response text lacks flexibility and struggles to meet users' personalized needs. Furthermore, some intelligent customer service systems assist human customer service in screening and answering basic inquiries. They can quickly find and collect relevant business answers using keywords, but they suffer from inaccurate semantic understanding of user questions, leading to imprecise responses. For example, incorrect keywords provided by the user will result in incorrect answers. Some systems also cannot customize responses based on users' individual needs; additionally, some systems perform poorly when handling complex scenarios. Summary of the Invention

[0004] In view of this, the present invention provides an intelligent customer service method and system to solve the aforementioned conventional problems.

[0005] This invention provides an intelligent customer service method comprising the following steps: receiving demand information and auxiliary information generated by a user in an interactive module; generating demand type data based on the demand information and user scenario data based on the auxiliary information; matching a first text group in a database module using vector indexing technology based on the demand type data and user scenario data; generating a response text using a natural language processing model based on the demand information and the first text group; and providing a reply to the user through the interactive module based on the response text.

[0006] The steps of generating requirement type data from requirement information and generating user scenario data based on auxiliary information further include the following steps: performing a first vectorization process on the requirement information to generate requirement type data; and performing a second vectorization process on the auxiliary information to generate user scenario data.

[0007] The process of generating a response text using a natural language processing model based on the demand information and the first text group further includes the following steps: inputting the demand information and the first text group into the natural language processing model for semantic understanding and contextual analysis; and generating a response text based on the result of the semantic understanding and contextual analysis, according to the response generation algorithm.

[0008] The answer generation algorithm includes one or more of rule-based algorithms, template-based algorithms, or generative models based on deep learning.

[0009] The answer generation algorithm further performs grammatical or semantic logic verification based on the results of semantic understanding and contextual analysis.

[0010] The intelligent customer service method further includes the following steps: after receiving a reply, the user adds a tag according to the type of need or the user scenario information in the reply; and the type of need information and the user scenario information with the tag information are stored in the database module.

[0011] The generation of response text using natural language processing models is also based on historical customer service Q&A data.

[0012] The auxiliary information includes one or more of the following: the user's historical interaction records, device or application status and operation information, contextual information, user preferences or personal information, and information from external data sources.

[0013] The interactive modules include one or more of the following interactive methods: text interaction, voice interaction, image interaction, video interaction, device or application operation interaction, virtual reality or augmented reality interaction.

[0014] This invention provides an intelligent customer service system, comprising an interaction module, a service module, and a database module. The interaction module is configured to interact with users, receiving and responding to their needs and supplementary information. The service module generates need type data based on the need information and user scenario data based on the supplementary information. The database module stores the need type data and user scenario data. The service module generates response text based on the need type data, user scenario data, and the data stored in the database module.

[0015] In summary, this invention provides an intelligent customer service method and system with the following advantages: The intelligent customer service method and system disclosed in this invention fully utilize technologies such as natural language processing, vector indexing, and artificial intelligence. Through an intelligent customer service system, it can quickly and accurately identify user needs and provide expected responses without waiting for human intervention, greatly shortening user waiting time and improving service efficiency. Through personalized responses and targeted solutions, the intelligent customer service system can better meet users' individual needs, enhancing user experience and increasing user satisfaction. The intelligent customer service system can continuously learn and optimize based on user feedback and interaction data, constantly improving service quality and accuracy. Through the intelligent customer service system, users can enjoy efficient and personalized customer service anytime, anywhere, and it has broad application prospects in the field of customer service. Simple Explanation of the Diagram

[0016] Figure 1 is a flowchart illustrating an embodiment of the intelligent customer service method of the present invention.

[0017] Figure 2 is a flowchart illustrating the implementation of an intelligent customer service method according to one embodiment of the present invention.

[0018] Figure 3 is a schematic flowchart illustrating the interactive information vectorization processing of an intelligent customer service method according to one embodiment of the present invention.

[0019] Figure 4 is an interactive schematic diagram illustrating the auxiliary information of a user in an intelligent customer service method according to one embodiment of the present invention.

[0020] Figure 5 is a schematic diagram illustrating the analysis of auxiliary information in an intelligent customer service method according to one embodiment of the present invention.

[0021] Figure 6 is an interactive schematic diagram illustrating the user's demand information in an intelligent customer service method according to one embodiment of the present invention.

[0022] Figure 7 is a schematic diagram illustrating the composition of an intelligent customer service system according to one embodiment of the present invention. Implementation

[0023] To make the advantages, spirit, and features of the present invention more easily and clearly understood, detailed descriptions and discussions will follow with reference to specific embodiments and the accompanying drawings. It is important to note that these specific embodiments are merely representative examples of the present invention, and the specific methods, systems, etc., exemplified are not intended to limit the present invention or the corresponding specific embodiments.

[0024] Please refer to Figure 1, which is a flowchart illustrating an embodiment of the intelligent customer service method of the present invention. In step S100, the intelligent customer service system can receive user demand information in various ways, including text input, voice input, image input, user-end logs, device operation records, etc. For example, users can describe their problems or needs by inputting text or voice in the interactive module, or they can upload pictures, videos, or user-end logs to provide relevant information. The demand type data and user scenario data are vectorized data, which can be converted from user demand information and auxiliary information after preprocessing and feature extraction. In actual use, firstly, the received user demand information and auxiliary information are preprocessed, including noise reduction, word segmentation, part-of-speech tagging, etc. Then, feature extraction is performed on each word or phrase, such as word vectorization, TF-IDF (term frequency-inverse document frequency) calculation, etc., to convert them into numerical features. Next, the numerical features obtained after preprocessing and feature extraction are converted into vector form. This can be achieved by mapping each word or phrase to a vector representation in a high-dimensional space, where each dimension of the vector corresponds to a feature. Through the above methods, the intelligent customer service system can effectively transform user needs and auxiliary information into vectorized data.

[0025] When receiving user request information, the system can better understand the user's current needs and preferences by analyzing the user's historical interaction records. This can be achieved through the following steps: First, the system retrieves and analyzes the user's past interaction records, including conversation content, click behavior, and purchase history. Then, based on these historical interaction records, the system identifies the user's preferences, habits, and behavioral patterns, and processes this information as supplementary information along with the currently received request information. During user interaction with the system, the system can also consider current contextual information to better understand the user's needs and intentions. Contextual information can include the user's current environment, previous conversation content, and current operating status. The system can acquire and understand current contextual information by monitoring and analyzing user actions, conversation content, and environmental changes. For example, the system can identify the type of device the user is currently using, their geographical location, and the current time, and adjust and optimize service content and methods accordingly. In addition to user-provided information and internal system data, the intelligent customer service system can also utilize external sources to enrich and improve service content. These external sources can include public databases, social media platforms, and industry data. For example, the system can access the APIs of social media platforms to obtain users' messages and activities on social media, thereby understanding users' interests, social circles, and other information, and thus better personalize service content recommendations.

[0026] Optionally, the method for generating demand type data and user scenario data based on demand information includes the following steps: Step S110: Perform a first vectorization process on the demand information to generate demand type data. In this embodiment, the first vectorization process is to use natural language processing technology to vectorize the demand information. First, the received demand information is preprocessed, including removing special characters, stop words, punctuation marks, etc., and performing operations such as stemming or lemmatization to reduce vocabulary redundancy and noise. Based on the preprocessed demand information, a vocabulary is constructed, mapping each word to a unique index value. Such a vocabulary can serve as a reference for subsequent vectorization processing. The preprocessed demand information is then vectorized. Finally, the vectorized demand information is saved as demand type data or transmitted to other modules of the system for further processing. In this way, the system can perform subsequent scene matching, message retrieval, and other operations based on the demand type data, thereby realizing the function of intelligent customer service. Vectorization of demand information can be achieved using techniques such as the Bag of Words (BoW) model or word embeddings, but this invention is not limited thereto.

[0027] Step S120: Perform a second vectorization process on the auxiliary information to generate user scenario data. In this embodiment, this user scenario data may include diverse information such as the user's text description, images or video content, and operation logs, to comprehensively reflect the user's needs and environmental conditions.

[0028] Optionally, the auxiliary information is subjected to a second vectorization process to generate user scenario data, including: Step S121: When the auxiliary information is text data, the auxiliary information is vectorized using natural language processing technology to generate user scenario data. In this embodiment, for log information, formatting processing, key information extraction, and other operations are required. Key features can be extracted and converted into vector form.

[0029] Step S122: When the auxiliary information is image or video data, feature extraction based on convolutional neural networks is performed on the auxiliary information, and vectorization processing is performed on the extracted features to generate user scene data. In this embodiment, for multimedia information such as images and videos, feature extraction and dimensionality reduction are required. Deep learning models such as convolutional neural networks (CNN) or recurrent neural networks (RNN) can be used to extract features and represent the extracted features in vector form.

[0030] In step S200, based on the demand type data and user scenario data, a first text group is matched in a database module using vector indexing technology. In this embodiment, the database module is defined as a vector database, a database system used to store and retrieve vector data. In the vector database, data is stored in the form of vectors, each vector representing a data entity or a feature. The vector database provides an efficient storage and retrieval mechanism, enabling users to quickly query and retrieve data based on the similarity between vectors. In this embodiment, the database module stores text group vectors corresponding to multiple text groups. Matching the demand type data and user scenario data in the database module includes: using a vector similarity calculation method to compare the demand type vector and user scenario vector with the text group vector, and selecting the most similar text group as the first text group (TextChunks) matching result. The vector similarity calculation method can be any algorithm that measures the similarity between two vectors, such as cosine similarity or Jaccard similarity coefficient, and this invention is not limited to this.

[0031] In step S300: Based on the demand information and the first text group, a natural language processing model is used to generate a response text. In this embodiment, the demand information and the first text group are concatenated into a Prompt, which is a text segment containing the user's demand information and the first text group, used as input into the natural language processing model for text generation. The concatenation can be done by simply linking the demand information and the first text group together, or by adding separators or other markers to distinguish different parts, depending on the specific situation.

[0032] In this embodiment, in addition to the demand information and the first text group, the natural language processing model can also incorporate historical customer service Q&A data as input to generate the response text. This historical data can come from actual customer service dialogue records, knowledge bases, FAQs (Frequently Asked Questions), and other sources.

[0033] The method in step 300 includes the following steps: Step S301: Input the demand information and the first text group into the natural language processing model for semantic understanding and contextual analysis. In this embodiment, after receiving the semantic vectors of the demand information and the first text group, the natural language processing model performs semantic understanding and contextual analysis. This includes inferring and understanding the semantic meaning of the text, as well as analyzing the logical relationships and contextual information between the texts. Through the neural network structure and training mechanism within the model, the model can capture the semantic similarity and logical connections between texts.

[0034] Step S302: Based on the results of semantic understanding and contextual analysis, the natural language processing model generates the answer text according to the answer generation algorithm. In this embodiment, the natural language processing model converts the results of semantic understanding and contextual analysis into an intermediate representation suitable for the answer generation algorithm. This may involve converting the output of semantic understanding into a specific data structure or representation so that subsequent algorithms can better understand and process it. Selecting a suitable answer generation algorithm according to the system's requirements and design can ensure the quality and efficiency of the generated answer text.

[0035] Optionally, the answer generation algorithm may include one or more of rule-based algorithms, template-based algorithms, or generative models based on deep learning.

[0036] In this embodiment, for rule-based or template-based algorithms, relevant content is filled in according to predefined rules or templates to generate answer text; for generative models based on deep learning, intermediate representations are used as input, and the model learns to generate natural and fluent text answers.

[0037] Optionally, the answer generation algorithm may also include grammatical or semantic logic verification of the results of semantic understanding and contextual analysis.

[0038] In this embodiment, firstly, the system performs grammatical verification on the generated response text to ensure it conforms to the basic grammatical rules of the language. This includes checking sentence structure, the correctness of parts of speech, and the logical relationships between sentence components. For example, the system can use language models or grammatical analysis tools to detect and correct potential grammatical errors or non-standard expressions in the sentences. Secondly, the system performs semantic logic verification to ensure the response text is semantically accurate and reasonable. This includes checking the logical structure, semantic coherence, and consistency of information in the response text. For example, the system can use logical reasoning or semantic analysis techniques to detect potential logical errors or semantic ambiguities in the response text and provide corresponding correction suggestions or improvement plans. By performing grammatical or semantic logic verification on the results of semantic understanding and contextual analysis, the intelligent customer service system can further improve the quality and reliability of the generated response text, ensuring that the responses received by users are accurate, clear, and logical.

[0039] Optionally, the natural language processing model may be one or more of ChatGPT, LangChain, or Gemini, but this invention is not limited thereto.

[0040] In step S400: Based on the response text, a reply is provided to the user through the interactive module. In this embodiment, the specific interaction method can be selected and switched according to the user's preferences and the device's support. For example, for text interaction, the system can display text messages on the user interface and provide a text input box for the user to input; for voice interaction, the system can convert the user's spoken input into text using voice recognition technology and reply to the user in voice or text form; for image interaction, the system can display image messages and allow the user to interact by clicking or dragging. After receiving the generated response text, the interactive module displays the response text to the user and provides corresponding interactive functions. The user can further ask questions, issue commands, or express needs based on the response content. The interactive module needs to implement a corresponding response mechanism to promptly handle the user's requests and generate corresponding replies or perform corresponding operations. For example, if the user needs to learn more about a certain topic, the interactive module can provide relevant links or detailed information based on the user's request; if the user needs to perform a certain operation, the interactive module can provide corresponding buttons or menus for the user to select.

[0041] Although not shown in Figure 1, optional intelligent customer service methods also include: after the user receives a reply, adding tags according to the user's reply type information or user scenario information;

[0042] In this embodiment, based on user behavior, the system can tag demand type data or user scenario data. These tags can include information such as response satisfaction, user preferences, areas of interest, and purchase intentions, helping the system better understand user needs and behavioral patterns, and providing more personalized services. After receiving a customer service response, users can provide feedback through likes, emojis, comments, etc. The intelligent customer service system collects user feedback and analyzes it. Based on the analysis results, the intelligent customer service system can tag demand type data or user scenario data to improve data accuracy and model performance. The tagged demand type data and user scenario data can be used for subsequent processing by the intelligent customer service system, such as customized recommendations and personalized responses. The system can adjust the response text generation strategy based on user tag information to provide response content that better meets user needs and preferences, thereby improving user satisfaction and experience quality. Through the above implementation methods, the intelligent customer service system can more accurately understand user needs and provide targeted personalized services, thereby enhancing the system's intelligence level and user satisfaction.

[0043] Store the demand type information and user scenario information with tags into the database module.

[0044] Optionally, the interaction methods of the interactive module include one or more of the following: text interaction, voice interaction, image interaction, video interaction, device or application operation interaction, virtual reality or augmented reality interaction.

[0045] In this embodiment, users can interact with the intelligent customer service system by entering text on the interactive interface. Users can also interact by uploading images or screenshots. The system can use computer vision technology to analyze and understand the uploaded images, extract relevant information, and generate corresponding response text or operation suggestions. Users can interact with the system through video calls or by uploading video files. The system can analyze and process the video or video content to identify the user's needs or problems and provide corresponding answers or solutions. Users can interact with the system by operating the device or application status. For example, by clicking buttons, touching the screen, or executing specific operation commands to send needs or requests to the system, the system will respond accordingly based on the received operation messages. Users can interact with the system using virtual reality devices or augmented reality technology. The system can obtain interactive information through user behavior and actions in virtual environments or augmented reality scenes and respond or provide guidance accordingly. In the intelligent customer service method of this embodiment, the sequence number of each step is only to distinguish the steps and is not a limitation on the specific execution order of the steps. The execution order between the above steps can be adjusted and changed as needed.

[0046] Furthermore, in Embodiment 1, please refer to Figure 2, which illustrates an implementation flowchart of an intelligent customer service method according to one embodiment of the present invention. The implementation flowchart schematically presents one set of steps for intelligent customer service to provide services to users. The implementation steps provided by the intelligent customer service method of the present invention are not limited to this; other implementation methods using the intelligent customer service method proposed in this invention also fall within the scope of protection of this invention.

[0047] Step P100: The user submits a request to the intelligent customer service. In this step, the intelligent customer service method receives the user's request information from the interaction module and generates request type data based on the request information. Requests in this step primarily include those submitted via text input. However, auxiliary information input via voice input, image input, user-end logs, device operation records, etc., is also possible.

[0048] Step P200: The intelligent customer service provides the first round of responses based on the demand index. In this step, the question is semantically broken down based on the user's question, and the user's behavioral background is obtained based on the auxiliary information recorded in the server log, resulting in the first round of responses.

[0049] In this step, please refer to Figure 3, which is a flowchart illustrating the interactive information vectorization processing of an embodiment of the intelligent customer service method of the present invention. The intelligent customer service method performs a first vectorization processing on the demand information to generate demand type data. A second vectorization processing is performed on the auxiliary information to generate user scenario data. Based on the demand type data and user scenario data, vector indexing technology is used to match the first text group in the database module. The first vectorization processing refers to the process of converting demand information into vector form. In the intelligent customer service system, natural language processing technology can be used to perform word segmentation, part-of-speech tagging, syntactic analysis, and semantic analysis on the demand information, and then convert the analysis results into vector form. Demand type data refers to vector data representing the type of user demand. Demand types can be divided into various categories, such as consultation, complaint, and suggestion. Demand type data helps the intelligent customer service system quickly identify the type of user demand and select an appropriate response template or generation algorithm. The second vectorization processing refers to the process of converting auxiliary information into vector form. Auxiliary information may include user historical interaction records, device or application status and operation information, contextual information, user preferences or personal information, and external data sources, etc. In intelligent customer service systems, various data processing techniques can be used to process auxiliary information, and the results are then converted into vector form. User scenario data refers to vector data representing the user's current scenario. User scenarios can include the user's location, device used, and operating environment. User scenario data helps intelligent customer service systems better understand user needs and provide more accurate responses.

[0050] Subsequently, based on the demand information and the first text set, a natural language processing (NLP) model is used to generate a response text. This includes inputting the demand information and the first text set into the NLP model for semantic understanding and contextual analysis; and based on the results of semantic understanding and contextual analysis, the NLP model generates a response text according to the response generation algorithm.

[0051] Step P300: Determine user satisfaction. During this step, the system performs a consistency check on the response, including grammatical or semantic logic verification. First, the system performs grammatical verification on the generated response text to ensure it conforms to basic grammatical rules. This includes checking sentence structure, part-of-speech correctness, and logical relationships between sentence components. For example, the system can use language models or grammatical analysis tools to detect and correct potential grammatical errors or non-standard expressions. Second, the system performs semantic logic verification to ensure the response text is semantically accurate and reasonable. This includes checking the logical structure, semantic coherence, and message consistency of the response text. For example, the system can use logical reasoning or semantic analysis techniques to detect potential logical errors or semantic ambiguities in the response text and provide corresponding correction suggestions or improvement plans. By performing grammatical or semantic logic verification on the results of semantic understanding and contextual analysis, the intelligent customer service system can further improve the quality and reliability of the generated response text, ensuring that the responses received by users are accurate, clear, and logical.

[0052] Step P400: A text response or action command can be generated based on the answer, and the user can then click on the app. If the customer is satisfied with the response in step P300, proceed to this step. In this step, the intelligent customer service method provides a response to the user through the interactive module based on the response text. In this embodiment, the specific interaction method can be selected and switched according to the user's preferences and device support. Two response formats can be generated based on the answer: 1. Text or verbal format; 2. The response can be converted into an action executed by the app.

[0053] Step P500: The user provides information to aid in the decision-making process. This step is initiated when the customer is dissatisfied with the response in step P300. In this step, the intelligent customer service method receives the user's requests and supplementary information from the interactive module. Users can describe their problems or needs by entering text or voice in the interactive module, or by uploading images, videos, user-side logs, or following the prompts of the intelligent customer service to provide relevant information.

[0054] Step P600: AI Analysis of User Question Supporting Information. In this step, AI analysis is further performed based on historical chat logs and user log records. First, the requirement information undergoes vectorization processing to generate requirement type data. Then, the supporting information undergoes second vectorization processing to generate user scenario data. This includes: when the supporting information is text data, natural language processing techniques are used to vectorize the supporting information to generate user scenario data; when the supporting information is image or video information, feature extraction based on convolutional neural networks is performed on the supporting information, and vectorization processing is performed based on the extracted features to generate user scenario data.

[0055] Step P700: Complete the question index response for this round. In this step, if the intelligent customer service method responds to user questions more than two times, each round of response uses vector indexing technology to match the Nth text group of the current round in the database module based on all demand type data and user scenario data of the current round and the aforementioned interactions.

[0056] Subsequently, based on the demand information and the Nth text group, a natural language processing model is used to generate a response text, including: inputting the demand information and the Nth text group into the natural language processing model for semantic understanding and contextual analysis; and generating a response text based on the results of semantic understanding and contextual analysis according to the response generation algorithm.

[0057] Step P800: Complete customer service and end the conversation. In this step, the customer service representative ends the conversation. If the intelligent customer service method is used, after the user receives the response, tags are added to the demand type information or user scenario information based on the user's further actions.

[0058] Step P900: End Service. In this step, the intelligent customer service method stores the demand type information and user scenario information with tagged messages into the database module.

[0059] In Embodiment 2, please refer to Figure 4, which is an interactive schematic diagram illustrating the auxiliary information for a user in an intelligent customer service method according to one embodiment of the present invention. For example, as shown in Figure 4, the intelligent customer service determines, based on system information, that the user's current operating interface is a blood pressure measurement interface and that the user has not performed any operation that day. In this scenario, the user clicks the microphone to ask the intelligent customer service about the blood pressure monitor's inability to measure. In addition, the intelligent customer service will also combine the user's purchase history or the model of the blood pressure monitor currently connected to the operating interface to obtain more comprehensive auxiliary information. This information includes the page the user is currently operating, the question the user has asked, and device information related to the question.

[0060] This embodiment uses a typical scenario where a user raises a device problem with the intelligent customer service to illustrate the interaction of user demand information in the intelligent customer service method of this embodiment.

[0061] Based on the user's request information, such as the user's complaint that the blood pressure monitor cannot measure, the intelligent customer service obtains the application status log (APP log) after obtaining the user's authorization. Users can also upload relevant information such as page screenshots as prompted by the intelligent customer service to further diagnose and resolve the problems encountered by the user.

[0062] Please refer to Figures 4 and 5 together. Figure 5 is a schematic diagram illustrating the analysis of auxiliary information in an embodiment of the intelligent customer service method of the present invention. As shown in Figure 5, in this embodiment, after receiving the user's request information and corresponding auxiliary information, including application status log (APP log) messages, the intelligent customer service needs to perform formatting processing and key message extraction, i.e., the first vectorization processing, on the request information to extract the request type data. In this embodiment, the request type is information-type messages. The auxiliary information is then formatted and key message extracted, i.e., the second vectorization processing, to obtain user scenario data. In this embodiment, the auxiliary information includes the application status log (APP log), and the obtained user scenario data includes key features extracted from the corresponding logs of the user device. The log messages uploaded as auxiliary information include date and time, log type, page name, and error messages, etc.

[0063] Subsequently, based on the user's requested information and supplementary information, the intelligent customer service uses vector indexing technology to match the first text group relevant to the user's current scenario in the database module. For example, in the log messages provided by the user, the intelligent customer service might match that the problem the user encountered was due to certain permissions required by the blood pressure monitor application not being enabled.

[0064] Next, based on the user's needs and the first matched text group, the intelligent customer service uses a natural language processing model to generate a response text for the user's question. The user will receive a reply from the intelligent customer service, which may include specific steps needed to resolve the issue, such as prompting the user to grant the appropriate Bluetooth permissions. Depending on the system settings, the intelligent customer service may also pop up a permission operation page to guide the user through further operations.

[0065] For example, even if a user has enabled Bluetooth, they may still be unable to connect to the device. The intelligent customer service may prompt the user via voice to generate an interface for the required location permissions and guide them to perform specific operations, such as clicking "Allow during use only" to enable Bluetooth location permissions.

[0066] After the user completes the corresponding operation according to the instructions of the intelligent customer service, the problem will be resolved and the user can use their device normally, thus achieving an effective solution to the user's problem.

[0067] If the customer is not satisfied, for example, if the user's problem is not resolved or there are other issues that require continued assistance, and multiple rounds of responses are needed, the intelligent customer service will continue to perform first and second vectorization processing based on all the user's feedback and by obtaining logs or providing corresponding auxiliary information by guiding the user's operation, to obtain the Nth text data for the Nth round of responses, and continue to provide responses until the customer service is completed.

[0068] Optionally, after completing each round or the entire service process, if the user finds the customer service content accurate and complete, they can give the response a "like." The intelligent customer service system collects user feedback and analyzes it along with user needs information and supplementary information. The analysis results show that users' needs for addressing the issues raised in this round are very common, and the existing response interaction methods can effectively meet user needs. Therefore, the intelligent customer service system can mark this type of need as a "high-frequency need" and store it in the database. Furthermore, the intelligent customer service system can continuously improve the response text based on user feedback. For example, if a user points out an error or omission in a comment, the intelligent customer service system can correct the response text and update the data in the database.

[0069] In Embodiment 3, please refer to Figure 6, which is an interactive schematic diagram illustrating the user's needs information in an intelligent customer service method according to one embodiment of the present invention. As shown in Figure 6, the intelligent customer service, based on system information, obtains that the user's current operating page interface is a blood pressure measurement interface, and analyzes and determines that the user has not yet performed any operation that day, as supplementary information. In addition, the intelligent customer service can also combine the user's purchase history or information such as the model of the blood pressure monitor connected to the user's operating interface as supplementary information. This background information helps the intelligent customer service to more accurately understand the user's needs and context, thereby providing more personalized and precise services.

[0070] This embodiment uses a typical scenario where a user requests operational guidance from an intelligent customer service representative to illustrate the interaction of user request information in the intelligent customer service method of this embodiment.

[0071] The user presses and holds the microphone button to ask how to measure blood pressure. This voice message is converted into text by the intelligent customer service system and sent as the user's request.

[0072] After receiving the demand information and auxiliary information, the intelligent customer service generates demand type data as consultation data. Based on the auxiliary information, it generates user scenario data, including the blood pressure monitor model. It obtains the first text group data through vector indexing and combines it with the demand information to generate response text, such as asking the user, "Is there a problem with the operation today?"

[0073] This interactive approach aims to further confirm user needs and understand their current operational status. After confirming the user's consent, the intelligent customer service will generate a corresponding response based on system information and user needs, and broadcast it in voice format according to the user's settings. The intelligent customer service's response may include operational instructions for blood pressure measurement or suggestions for resolving problems. Subsequently, the intelligent customer service will continue to guide the user to operate according to the generated response, providing operational guidance based on the user's set interaction method, such as voice. This flexibility of interaction allows the intelligent customer service to provide personalized services and operational guidance based on the user's actual needs and preferences.

[0074] This embodiment demonstrates an interactive method in which intelligent customer service provides services to users by following the user's usage scenario. It does not perform any preset actions, but provides interactive guidance to users based on their operations in order to provide intelligent customer service.

[0075] Please refer to Figure 7, which is a schematic diagram illustrating the composition of an intelligent customer service system according to one embodiment of the present invention. As shown in Figure 7, this embodiment of the present invention provides an intelligent customer service system 1 for implementing the above-described intelligent customer service method. The intelligent customer service system includes an interaction module M100, a service module M200, and a database module M300. The interaction module M100 is configured to interact with a user, receiving and responding to the user's request information and auxiliary information. The service module M200 is configured to generate request type data based on the request information, and generate response text based on the request type data, user scenario data, and data from the database module M300. The database module M300 stores the request type data and user scenario data. In this embodiment, the database module M300 is configured as a vector database, which can effectively store and process vector data.

[0076] Optionally, the M300 database module also includes different types of databases, such as relational databases and NoSQL databases, to expand its applicability and functionality. Relational databases can be used to store structured data, while NoSQL databases can be used to store unstructured data, supporting more diverse data storage needs. This design can provide intelligent customer service systems with more flexible and comprehensive data management and query functions, thereby improving system performance and efficiency.

[0077] The detailed description of the preferred embodiments above is intended to more clearly illustrate the features and spirit of the present invention, and is not intended to limit the scope of the invention to the preferred embodiments disclosed above. Rather, the aim is to cover various modifications and equivalent arrangements within the scope of the patent claims made by this invention. Therefore, the scope of the patent claims made by this invention should be interpreted in the broadest possible sense based on the foregoing description, so as to cover all possible modifications and equivalent arrangements.

[0078] S100, S200, S300, S400: Steps

[0079] P100, P200, P300, P400, P500, P600, P700, P800, P900: Steps

[0080] 1: Intelligent Customer Service System

[0081] M100: Interactive Module

[0082] M200: Service Module

[0083] M300: Database Module

Claims

1. An intelligent customer service method, comprising the following steps: a service module receives a user's demand information and auxiliary information generated in an interaction module; the service module generates demand type data based on the demand information and user scenario data based on the auxiliary information; the service module matches a first text group in a database module using a vector indexing technique based on the demand type data and the user scenario data; the service module generates a response text using a natural language processing model based on the demand information and the first text group; and the service module provides a reply to the user through the interaction module based on the response text.

2. The intelligent customer service method as described in claim 1, wherein the steps of the service module generating the demand type data based on the demand information and generating the user scenario data based on the auxiliary information further include the following steps: the service module performs a first vectorization processing on the demand information to generate the demand type data, wherein the first vectorization processing employs natural language processing technology; and the service module performs a second vectorization processing on the auxiliary information to generate the user scenario data, wherein the second vectorization processing employs one of natural language processing technology and deep learning technology.

3. The intelligent customer service method as described in claim 1, wherein the service module generates the response text using a natural language processing model based on the demand information and the first text group, further comprising the following steps: the service module inputs the demand information and the first text group into the natural language processing model for semantic understanding and contextual analysis; and the natural language processing model generates the response text according to a response generation algorithm based on a result of semantic understanding and contextual analysis.

4. The intelligent customer service method as described in claim 3, wherein the answer generation algorithm is any combination of a rule-based algorithm, a template-based algorithm, and a generative model based on deep learning.

5. The intelligent customer service method as described in claim 3, wherein the answer generation algorithm further performs grammatical or semantic logic verification through the results of a pair of semantic understanding and contextual analysis.

6. The intelligent customer service method as described in claim 1 further includes the following steps: the service module receives feedback from the user in response to the reply, and adds a tag to the demand type data or the user scenario data according to the feedback; and the service module stores the demand type data and the user scenario data with one of the tag information into the database module.

7. The intelligent customer service method as described in claim 1, wherein the service module uses a natural language processing model to generate the answer text based on historical customer service Q&A data.

8. The intelligent customer service method as described in claim 1, wherein the auxiliary information includes one or more of the following: historical interaction records of one of the users, the status of a device or an application, operation information, contextual information, user preferences or personal information, and information from an external data source.

9. The intelligent customer service method as described in claim 1, wherein the interaction mode of the interactive module includes one or more of text interaction, voice interaction, image interaction, video interaction, operation interaction of device or application, virtual reality or augmented reality interaction.

10. An intelligent customer service system, comprising: an interaction module for setting up interaction with a user, receiving one of the user's needs and auxiliary information; a service module for generating need type data based on the need information and generating user scenario data based on the auxiliary information; and a database module for storing the need type data and the user scenario data; wherein, The service module generates a response text based on the demand type information, the user scenario information, and the data stored in the database module, and the interaction module provides the response text to the user.

Images