Intelligent power customer service method and related equipment

Abstract

The invention provides an intelligent power customer service method and related equipment, and the method comprises the steps: obtaining first user input information which comprises at least one of voice data, text data or video data; performing semantic analysis on the first user input information to determine first user service information; the first user service information is used for indicating a service task; matching a corresponding service operation process based on the first user service information, wherein the service operation process comprises a directed graph formed by a plurality of service operation nodes; selecting a corresponding path along the directed graph based on node input information corresponding to the service operation node so as to execute the service task; the node input information is obtained by performing semantic analysis on second user input information of a user for the service operation node; and returning an execution result of the service task to the user.

Description

Technical Field

[0001] This disclosure relates to the power sector, and more particularly to a smart power customer service method and related equipment. Background Technology

[0002] With the deepening of digital transformation in the power industry, customer service, as a crucial bridge between power companies and users, directly impacts customer satisfaction and the company's market competitiveness. Traditional power customer service systems primarily rely on human agents or rule-based question-and-answer systems for service responses, resulting in low efficiency, slow response times, and insufficient professionalism. In recent years, the development of artificial intelligence technology, particularly the breakthroughs in natural language understanding and generation using Large Language Models (LLMs), has provided new possibilities for upgrading intelligent customer service systems.

[0003] However, applying large language models to power customer service scenarios still faces many challenges. First, the power industry is highly specialized and complex, involving multiple sub-sectors such as electricity consultation, electricity bill inquiry, fault reporting, and business processing, placing higher demands on the accuracy of customer intent recognition. Second, customer needs are diverse and expressed in inconsistent ways, making traditional single-intent recognition methods insufficient to meet practical needs. Furthermore, large language models may exhibit "illusion" phenomena when generating responses, meaning the generated content deviates from facts or business data, potentially misleading users or even incurring legal risks. Therefore, there is an urgent need for an intelligent power customer service method that can accurately identify customer intent, effectively match standardized service processes, and achieve efficient responses while ensuring business accuracy.

[0004] The following technical problems exist in existing intelligent customer service systems for the power industry: Inaccurate customer intent identification: When faced with diverse customer expressions, the existing system struggles to accurately identify the customer's true intent, leading to incorrect service paths or missing information.

[0005] Non-standardized service processes: Traditional customer service systems lack standardized service chain design and cannot dynamically adjust service processes according to customer intentions.

[0006] The application of large language models carries a high risk of illusion: when generating responses directly using large language models, content that does not match the actual business data is likely to appear, affecting the credibility and security of the service.

[0007] Weak contextual understanding: The lack of an effective contextual modeling mechanism in multi-turn dialogues affects the coherence of the interaction and the user experience.

[0008] Difficulty in integrating business data: There is a lack of an effective connection mechanism with the power business platform, making it impossible to obtain and integrate customer-related business data in real time. Summary of the Invention

[0009] In view of this, the purpose of this disclosure is to propose a smart power customer service method and related equipment.

[0010] In a first aspect, this disclosure provides a method for providing intelligent electricity customer service, comprising: Obtain first user input information, wherein the first user input information includes at least one of voice data, text data, or video data; Semantic analysis is performed on the first user input information to determine the first user service information; the first user service information is used to indicate service tasks. Based on the first user service information, a corresponding service operation process is matched, and the service operation process includes a directed graph formed by multiple service operation nodes; Based on the node input information corresponding to the service operation node, a corresponding path is selected along the directed graph to execute the service task; the node input information is obtained by semantic analysis based on the second user input information of the user for the service operation node; The execution result of the service task will be returned to the user.

[0011] A second aspect of this disclosure provides an intelligent electricity customer service device, comprising: The acquisition module is used to acquire first user input information, which includes at least one of voice data, text data, or video data. The semantic analysis module is used to perform semantic analysis on the first user input information to determine the first user service information; the first user service information is used to indicate service tasks. The process matching module is used to match the corresponding service operation process based on the first user service information. The service operation process includes a directed graph formed by multiple service operation nodes. The process execution module is used to select a corresponding path along the directed graph based on the node input information corresponding to the service operation node in order to execute the service task; the node input information is obtained by semantic analysis based on the second user input information of the user for the service operation node; The result return module is used to return the execution result of the service task to the user.

[0012] A third aspect of this disclosure provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in the first aspect.

[0013] In a fourth aspect, this disclosure provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method described in the first aspect.

[0014] A fifth aspect of this disclosure provides a computer program product including computer program instructions that, when executed on a computer, cause the computer to perform the method described in the first aspect.

[0015] As described above, this disclosure provides an intelligent power customer service method and related equipment. By acquiring information input from users in the form of voice, text, or video, and clarifying the service task through semantic analysis, a service flow containing multiple service operation nodes is matched. During execution, based on the node input information obtained from further analysis of user input at each node, the task is advanced along the directed graph, and the execution result is finally fed back to the user. This method can comprehensively accept diverse user input, accurately grasp service needs, and efficiently execute service tasks and provide timely feedback through an orderly service operation process and flexible path selection, effectively improving the quality and efficiency of power customer service and enhancing the user experience.

[0016] Furthermore, by establishing an intent service chain with structured representation, flow control logic, exception handling, and dynamic expansion, and a collaborative working mechanism between the intent service chain and the large language model, the natural language generation capabilities of the large language model are preserved while ensuring the standardization and reliability of the service process. Firstly, customer intent for service requests in the power customer service channel is identified and matched with the intent service chain, solving the problems of high professionalism in power customer service, diverse customer requests, and high risk of model illusion.

[0017] Through multi-round dialogue and contextual understanding, customer intent is accurately identified and matched with a standardized service chain to form a dynamic service script response string. Based on the requirements of the response string and customer intent, customer business data is extracted from the power customer service business platform. At the same time, the large language model organizes the script into a response message and sends it back to the customer, achieving efficient, accurate and secure power customer service, and significantly improving customer satisfaction and business processing efficiency.

[0018] This disclosure combines a large language model intent recognition method with standardized service chain matching to construct an intent service chain for power customer service, ensuring that customer requests receive a safe, accurate, and efficient response. This method utilizes the semantic understanding capabilities of a large language model to identify customer intent, while simultaneously ensuring the standardization and security of the service process through a predefined service chain. This effectively avoids the risks associated with model illusions, providing a safe, professional, and reliable solution for power customer service. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this disclosure or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a schematic diagram of a smart power customer service architecture according to an embodiment of this disclosure.

[0021] Figure 2 This is a schematic diagram of the structure of an exemplary electronic device according to an embodiment of the present disclosure.

[0022] Figure 3 A schematic flowchart illustrating a smart power customer service method according to an embodiment of this disclosure.

[0023] Figure 4 This is a schematic diagram of an intelligent power customer service system according to an embodiment of the present disclosure.

[0024] Figure 5 This is a schematic diagram of a smart power customer service device according to an embodiment of the present disclosure. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this disclosure clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0026] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this disclosure should have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms "first," "second," and similar terms used in the embodiments of this disclosure do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0027] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0028] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0029] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0030] Figure 1 A schematic diagram of a smart electricity customer service architecture according to an embodiment of this disclosure is shown. (See reference...) Figure 1 The intelligent power customer service architecture 100 may include a server 110, a terminal 120, and a network 130 providing communication links. The server 110 and the terminal 120 can be connected via a wired or wireless network 130. The server 110 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, security services, and CDN.

[0031] Terminal 120 can be implemented in hardware or software. For example, when terminal 120 is implemented in hardware, it can be various electronic devices with a display screen and support page display, including but not limited to smartphones, tablets, e-book readers, laptops, and desktop computers. When terminal 120 is implemented in software, it can be installed in the electronic devices listed above; it can be implemented as multiple software programs or software modules (e.g., software programs or software modules used to provide distributed services) or as a single software program or software module, without specific limitations.

[0032] It should be noted that the intelligent power customer service method provided in this embodiment can be executed by the terminal 120 or by the server 110. It should be understood that... Figure 1 The number of terminals, networks, and servers shown is for illustrative purposes only and is not intended to be a limitation. Any number of terminals, networks, and servers can be used depending on implementation needs.

[0033] Figure 2 A schematic diagram of the hardware structure of an exemplary electronic device 200 provided in an embodiment of this disclosure is shown. Figure 2As shown, the electronic device 200 may include: a processor 202, a memory 204, a network module 206, a peripheral interface 208, and a bus 210. The processor 202, memory 204, network module 206, and peripheral interface 208 are interconnected within the electronic device 200 via the bus 210.

[0034] Processor 202 may be a Central Processing Unit (CPU), a Neural Processing Unit (NPU), a Microcontroller (MCU), a programmable logic device, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits. Processor 202 can be used to perform functions related to the techniques described in this disclosure. In some embodiments, processor 202 may also include multiple processors integrated as a single logic component. For example, such as... Figure 2 As shown, processor 202 may include multiple processors 202a, 202b and 202c.

[0035] Memory 204 can be configured to store data (e.g., instructions, computer code, etc.). Figure 2 As shown, the data stored in memory 204 may include program instructions (e.g., program instructions for implementing the smart electricity customer service method of this disclosure) and data to be processed (e.g., the memory may store configuration files of other modules, etc.). Processor 202 may also access the program instructions and data stored in memory 204 and execute the program instructions to operate on the data to be processed. Memory 204 may include volatile storage devices or non-volatile storage devices. In some embodiments, memory 204 may include random access memory (RAM), read-only memory (ROM), optical disk, magnetic disk, hard disk, solid-state drive (SSD), flash memory, memory stick, etc.

[0036] Network module 206 can be configured to provide communication with other external devices to electronic device 200 via a network. This network can be any wired or wireless network capable of transmitting and receiving data. For example, the network can be a wired network, a local wireless network (e.g., Bluetooth, WiFi, Near Field Communication (NFC), etc.), a cellular network, the Internet, or a combination thereof. It is understood that the type of network is not limited to the specific examples described above. In some embodiments, network module 206 may include any combination of any number of network interface controllers (NICs), radio frequency modules, transceivers, modems, routers, gateways, adapters, cellular network chips, etc.

[0037] The peripheral interface 208 can be configured to connect the electronic device 200 to one or more peripheral devices to enable information input and output. For example, peripheral devices may include input devices such as keyboards, mice, touchpads, touch screens, microphones, and various sensors, as well as output devices such as displays, speakers, vibrators, and indicator lights.

[0038] Bus 210 can be configured to transmit information between various components of electronic device 200 (e.g., processor 202, memory 204, network module 206, and peripheral interface 208), such as internal buses (e.g., processor-memory bus), external buses (USB port, PCI-E bus), etc.

[0039] It should be noted that although the architecture of the above-described electronic device 200 only shows the processor 202, memory 204, network module 206, peripheral interface 208, and bus 210, in specific implementations, the architecture of the electronic device 200 may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the architecture of the above-described electronic device 200 may only include the components necessary for implementing the embodiments of this disclosure, and does not necessarily include all the components shown in the figures.

[0040] In the power industry, with the rapid development of new power businesses such as smart grids, distributed energy, and electric vehicles, customer demands for power services are becoming increasingly diversified and complex. Power customer service not only involves traditional basic services such as electricity bill inquiries and fault reporting, but also covers emerging fields such as renewable energy access consulting, energy efficiency management advice, and electricity market transactions. These service demands are often characterized by high professionalism, high context dependence, and significant personalization requirements, placing higher demands on the intelligence level of power customer service systems.

[0041] Traditional power customer service systems primarily rely on pre-defined rule bases, keyword matching, or simple machine learning models to handle customer inquiries. While these methods perform adequately with simple, standardized questions, they often struggle to accurately capture the customer's true intent in complex, multi-turn dialogue scenarios, leading to inaccurate responses and inefficiency. Furthermore, the power industry involves a large amount of technical terminology and business logic, and traditional methods have significant limitations in understanding and processing this complex information.

[0042] In recent years, Natural Language Processing (NLP) technology has made significant progress, especially with the rise of large language models (such as the GPT series and BERT), which has brought revolutionary changes to the field of intelligent customer service. Large language models possess powerful language understanding, contextual understanding, and generation capabilities, enabling them to better handle complex dialogue scenarios and provide a more natural and fluent interactive experience. However, in related technologies, power intelligent customer service systems based on traditional NLP technology often utilize keyword matching, rule engines, or simple machine learning models (such as Naive Bayes and Support Vector Machines) to identify the intent in customer input. These methods perform reasonably well when handling simple, standardized intents, but their accuracy is low when facing complex, multi-turn dialogue scenarios. Based on the identified intent, the system searches and matches corresponding service processes or answer templates from a pre-set service resource library. These service resources are usually static and difficult to adapt to diverse customer needs and changes in the business environment. Based on the matched service process or answer template, a response script is generated and returned to the customer. The response script often lacks personalization and contextual relevance, making it difficult to provide a high-quality customer service experience. Subsystems may provide interfaces with the power customer service business platform to extract customer-related business data. However, this data is usually only used for simple information display or verification and is not fully integrated into the response generation process of intelligent customer service.

[0043] Therefore, the limitations of this related technology are mainly reflected in the following aspects: Inaccurate intent recognition: Traditional methods struggle to handle complex, multi-turn dialogue scenarios, resulting in low accuracy in intent recognition.

[0044] Inflexible service resource matching: The static service resource library is difficult to adapt to the diverse needs of customers and changes in the business environment.

[0045] Lack of personalization and contextual relevance in responses: Responses often lack personalization and contextual relevance, making it difficult to provide a high-quality customer service experience.

[0046] The risk of model illusion has not been effectively controlled: Traditional methods have not fully considered the risk of model illusion, which may lead to inaccurate or misleading responses.

[0047] Therefore, how to accurately identify customer intent, efficiently match service resources, and ensure the accuracy and security of responses in power customer service has become an urgent technical problem to be solved.

[0048] See Figure 3 , Figure 3 A schematic flowchart of a smart electricity customer service method according to an embodiment of the present disclosure is shown. The smart electricity customer service method according to an embodiment of the present disclosure can be deployed on a server. Figure 3In this context, the intelligent power customer service method 300 may further include the following steps.

[0049] In step S310, first user input information is obtained, which includes at least one of voice data, text data, or video data.

[0050] The first user input information may refer to the initial information provided by the user to the intelligent power customer service system, and its form may include at least one of voice data (such as a recording of the user's request expressed by telephone), text data (such as the text content entered by the user in the online customer service chat box), or video data (such as a video taken by the user about a power equipment failure).

[0051] In step S320, semantic analysis is performed on the first user input information to determine the first user service information; the first user service information is used to indicate service tasks.

[0052] Semantic analysis refers to understanding and parsing user input information (including first user input and node input information formed based on second user input), identifying keywords, semantic relationships, etc., to extract useful content and intent. First user service information refers to the result obtained after semantic analysis of the first user input information, which clarifies the service tasks required by the user, such as specific service needs like inquiring about electricity bill payment methods or reporting power outages.

[0053] In some embodiments, semantic analysis is performed on the first user input information to determine first user service information, including: The first user input information is processed to obtain intermediate input information; Semantic features are extracted based on the intermediate input information to obtain intermediate input features; Preliminary user service information is determined based on the intermediate input features; The preliminary user service information is verified based on the preset service information; Upon successful verification of the preliminary user service information, the preliminary user service information is determined to be the first user service information. If the verification of the preliminary user service information fails, the preliminary user service information is updated based on the preset service information to obtain the first user service information.

[0054] In determining the first user service information through semantic analysis, the input data is first processed to obtain intermediate input information. Then, its semantic features are extracted to obtain intermediate input features, thereby determining preliminary user service information. Next, it is verified against preset service information. If verification is successful, the information is directly determined; otherwise, it is updated based on the preset service information to obtain the final first user service information. This approach fully utilizes data processing and feature extraction technologies, and leverages the preset information verification and update mechanism to effectively improve the accuracy and reliability of user service information determination. This ensures that subsequent service tasks are based on accurate information, thereby enhancing the quality of smart power customer service.

[0055] Specifically, semantic analysis of the first user input information can be performed based on a large language model to obtain the identification result of the user service task. This result is then matched and verified against the service intent in the power knowledge base, and potential identification errors are corrected to ensure that the intent identification complies with power supply service specifications. The service intent verification module, which can be a component of the power customer service system, can receive the identification result of the first user input information from the large language model and use information from the power knowledge base for verification and correction. This improves the accuracy of user service identification and ensures that service responses comply with the specifications and requirements of the power industry.

[0056] Pre-set service information can be found in the power knowledge base. The power knowledge base collects and organizes information such as power business terminology, process specifications, policies, and regulations, removing duplicate and invalid information to ensure data accuracy and consistency. The processed information is stored based on a reasonable knowledge base structure, such as a tree or graph structure, for rapid retrieval and matching. Furthermore, the power knowledge base can be updated regularly to reflect the latest business changes and policies, and a knowledge base maintenance mechanism can be established to ensure data timeliness and accuracy.

[0057] The service intent verification module receives the service identification results (i.e., first user service information) from the large language model based on the first user's input information, including the service type and possible related information. Based on the service identification results from the large language model, it retrieves relevant business terms, process specifications, policies, and regulations from the power knowledge base. For example, it matches and verifies the service identification results with the retrieved information to determine if the identification results conform to the specifications and requirements in the power knowledge base. If the service identification results from the large language model are found to be inconsistent with the information in the power knowledge base, the module will correct the identification error. Correction methods may include adjusting the intent type based on the information in the power knowledge base, supplementing or correcting related information, etc. The verified service identification results are then matched against the service operation process.

[0058] A rule engine can be designed to define the rules and logic for intent verification, such as formulating verification rules based on power business terminology, process specifications, policies, and regulations. The rule engine can automatically verify and correct the recognition results of the large language model based on the verification rules. Furthermore, it can continuously optimize the intent recognition capabilities of the large language model based on feedback from the service intent verification module, improving the accuracy and robustness of the recognition.

[0059] Therefore, the service intent verification module receives the first user service information identified by the large language model. Then, relying on a complete and real-time updated power knowledge base, which includes accurate information such as power business terminology, process specifications, policies and regulations, it uses natural language processing technology to perform semantic matching and analysis verification between the recognition results and the knowledge base content. Through the set rule logic or with the help of machine learning algorithms, it judges whether the intent identified by the large language model is accurate and compliant. Once a deviation or error is found, the recognition results are corrected and adjusted according to the correct information in the knowledge base. Finally, it outputs an intent verification result that conforms to power supply service specifications and is accurate and reliable, thereby ensuring that subsequent service responses can accurately meet user needs and industry requirements.

[0060] In some embodiments, it also includes at least one of the following: In response to the inability to determine the first user service information based on the first user input information, guidance information is generated to guide the user to input supplementary information; Based on the supplementary user input information and the first user input information, semantic analysis is performed to determine the first user service information; In response to the inability to determine the first user service information based on the supplementary user input information and the first user input information, a service anomaly is determined. Based on the exception type and service exception policy of the service exception, the corresponding exception handling operation is executed.

[0061] In this system, if the service information for the first user cannot be determined based on the initial user input, guidance information is generated to prompt the user to input supplementary information. Semantic analysis is then performed by combining the supplementary information with the original information to determine the service information. If the information still cannot be determined after this process, it is considered a service anomaly, and corresponding anomaly handling operations are executed according to the established strategy based on the anomaly type. This solution increases the likelihood of obtaining accurate service information by guiding users to supplement their information, ensuring the smooth progress of the service process. The anomaly determination and handling mechanism effectively addresses various uncertainties, avoids service delays, improves the stability and reliability of smart power customer service, and optimizes the user experience.

[0062] In step S330, a corresponding service operation process is matched based on the first user service information. The service operation process includes a directed graph formed by multiple service operation nodes.

[0063] The service operation process can be a specific flow for executing service tasks determined based on the first user's service information, and is composed of a directed graph formed by multiple service operation nodes. The directed graph specifies the sequence and logical relationships between the service operation nodes, ensuring that service tasks can be executed in an orderly manner according to certain steps. A service operation node can be the basic unit in the service operation process, with each node representing a specific operation step or decision point. For example, in the service operation process for handling user-reported repair faults, there may be different service operation nodes such as "fault type judgment," "work dispatch," and "repair feedback."

[0064] In step S340, a corresponding path is selected along the directed graph based on the node input information corresponding to the service operation node to execute the service task; the node input information is obtained by semantic analysis based on the second user input information of the user for the service operation node.

[0065] The second user input information refers to further information provided by the user at a specific service operation node during the service operation process. This information can be in the form of voice, text, or video, used to supplement or provide feedback on relevant content at a specific operation node, enabling the system to execute service tasks more accurately.

[0066] In some embodiments, the node input information is obtained by semantic analysis based on second user input information from the user regarding the service operation node, including: The node service information corresponding to the service operation node is output to the user; the node service information is used to prompt the user for the information required to perform the service operation of the service operation node. Receive the second user input information from the user regarding the node service information; Semantic analysis is performed on the second user input information to obtain the node input information.

[0067] In this approach, when acquiring node input information, the system first outputs the node service information corresponding to the service operation node to the user, clearly prompting the user of the content required to perform the node operation and guiding the user to input information in a targeted manner. Next, it receives secondary user input information based on this feedback. Finally, it performs semantic analysis on this information to obtain the node input information. This solution, through clear prompts, makes user input more accurate and effective, reduces interference from invalid information, improves the accuracy and efficiency of semantic analysis in obtaining node input information, ensures the smooth progress of service operation nodes, and thus enhances the smoothness and quality of the entire smart power customer service process.

[0068] In some embodiments, the node service information corresponding to the service operation node is output to the user, including: Determine the user status based on the first user input information and / or the second user input information; In response to the user's state not meeting the state conditions, update the content and / or the number of times the node service information is updated; And / or, The node service information is generated based on the user's historical interaction information, the context information of the current interaction, and the node information template corresponding to the service operation node.

[0069] When outputting node service information corresponding to a service operation node to the user, the system determines the user's status based on the input information of the first and / or second user. If the user's status does not meet the preset conditions, the content or number of times the node service information is output is flexibly updated to better adapt to the user's situation. Furthermore, it generates node service information by comprehensively considering the user's historical interaction information, the current interaction context, and the node information template of the service operation node. This solution can dynamically and accurately generate and adjust node service information based on multiple dimensions such as the user's real-time status and past interactions, providing users with more targeted and practical guidance, effectively improving the user's interaction experience with the smart power customer service system, and ensuring the smooth progress of the service process.

[0070] In some embodiments, selecting a corresponding path along the directed graph based on the node input information corresponding to the service operation node includes: In response to detecting that the node input information satisfies the first condition of the service operation node, a first path associated with the service operation node in the directed graph is selected; In response to detecting that the node input information satisfies the second condition of the service operation node, a second path associated with the service operation node in the directed graph is selected; In response to the detection that the node input information does not meet any of the conditions of the service operation node, a service anomaly is determined.

[0071] In this scheme, when selecting a path along a directed graph based on the node input information of a service operation node, the path direction is determined by checking whether the node input information meets different preset conditions of the service operation node: if the first condition is met, the associated first path is selected; if the second condition is met, the associated second path is selected; if neither condition is met, the service is deemed abnormal. This scheme, based on clear conditional judgments, can accurately and efficiently determine the appropriate path in the directed graph according to the node input information, ensuring that the service operation process proceeds in the correct direction. Simultaneously, it promptly identifies abnormal situations, helping to improve the standardization and stability of smart power customer service processes and optimize service execution.

[0072] In some embodiments, a corresponding service processor is invoked based on the service operation interface corresponding to the service operation node to process the node input information and execute the service operation corresponding to the service operation node.

[0073] In this system, when executing a service operation corresponding to a service operation node, the system calls the matching service processor based on the pre-configured service operation interface of the node. The service processor then specifically processes the node's input information to complete the required service operation. This solution, through a clear mechanism of matching service operation interfaces with service processors, achieves professional and precise processing of input information from different service operation nodes. This ensures that each service operation node can efficiently and accurately execute its tasks, improving the overall operational efficiency and service quality of the intelligent power customer service system.

[0074] In step S350, the execution result of the service task is returned to the user.

[0075] The execution result of a service task can refer to the outcome produced by the system after completing the service task according to the service operation process, such as successfully answering a user's electricity bill question or repairing a power outage. The system will return this result to the user so that the user can understand the processing status of the service request.

[0076] See Figure 4 , Figure 4 A schematic diagram of an intelligent power customer service system according to an embodiment of the present disclosure is shown. Figure 4 In China, the intelligent power customer service system achieves automatic identification, processing, and feedback of customer issues through multi-module collaboration. Specifically, it includes modules such as customer interaction layer, intent recognition, service chain scheduling, business processing, and transfer to human customer service, which improves the efficiency and quality of power supply service customer interaction and provides power users with more convenient, efficient, and accurate power services.

[0077] The customer interaction layer can utilize various channels, such as web pages, mobile applications, and telephone customer service systems, for customer interaction with the smart power customer service system. Customers input their inquiries and problem descriptions through this layer, while simultaneously receiving processing results and suggestions from the system. The customer interaction layer is responsible for passing the customer's input information to subsequent intent guidance and recognition modules, and providing feedback on the response results to the customer. The interaction layer maintains multi-turn dialogue history through dialogue state tracking, and also stores customer interaction information in the customer interaction history in a uniquely identifiable way for subsequent determination of customer intent.

[0078] In customer service applications, the customer interaction layer and environmental perception are handled by intelligent customer service and online customer service, which can be considered existing intelligent agents. These agents perform customer service tasks for specific electricity customer service channels and can be viewed as environmental sensors. The customer interaction layer receives questions and answers from the environmental sensors, communicates with them, and records their historical interaction information.

[0079] The main intent guidance and recognition module combines intelligent algorithms from a large language model with predefined intent recognition rules. It processes the contextual information input by the customer at the interaction layer, along with the customer's interaction history, to guide the customer to explicitly express their intent and attempts to identify the customer's primary intent. If intent recognition is successful, the intent information is passed to the intent service chain scheduling center; if recognition fails, the information is passed to the intent guidance generation module. Simultaneously, it analyzes the emotional characteristics of the customer's dialogue in real time based on the latest and historical dialogue features (including speech rate, tone, and keywords) to detect the customer's emotions.

[0080] The intent guidance message generation module is based on natural language generation technology using a large-scale language model, possessing rich knowledge of the power industry and powerful language processing capabilities. When the main intent guidance and recognition module fails to identify the customer's intent, this module generates guidance messages based on the customer's input information, guiding the customer to further clarify their intent. The generated guidance messages are then fed back to the customer through the customer interaction layer. This provides core self-language and power customer language processing support for the entire intelligent power customer service system, including semantic understanding and knowledge retrieval in stages such as power customer intent recognition, guidance message generation, and intent processing.

[0081] The Intent Service Chain Scheduling Center employs a scheduling algorithm and rule engine. Once the main intent guidance and recognition module successfully identifies the customer's intent, the Intent Service Chain Scheduling Center schedules the corresponding intent processing unit for processing based on the intent type and preset rules. Simultaneously, it coordinates the workflow between different intent processing units. This mechanism transforms the power industry customer service process into a structured service chain directed graph. Through node type definition, execution condition judgment, jump control mechanism, and exception handling branches, it constrains and guides the output of the large language model, ensuring the professionalism and accuracy of power industry customer service.

[0082] The intent processing unit includes multiple processing modules tailored to different power business scenarios, such as an electricity bill inquiry processing unit, a fault reporting processing unit, and a business expansion processing unit. The intent service chain is modeled using a directed graph data structure, where each node represents a specific operation in the service process, and edges represent the execution order between nodes. In the directed graph model, nodes correspond to the basic operation units of the power business process, and edges represent process control relationships, forming a coherent sequence of operations. By establishing adjacency chains in the directed graph, the connection relationships between nodes in the intent chain can be clearly represented. This representation allows the process control logic of the intent chain to be presented intuitively.

[0083] When the intent processing unit receives an intent scheduling instruction, it first checks the cache for the intent processing chain. After a successful match, the intent processing unit, following the business processing instructions from the intent service chain scheduling center, calls the business processor to perform specific business logic and query the knowledge base, completing the collection of detailed information regarding the customer's power business-related requirements. Based on the frequency of intent chain usage, the intent processing unit stores the information in the intent service chain's cache, which resides in memory. For example, for knowledge-based information, it combines the power knowledge base to generate accurate answers. The process is as follows: Retrieval Phase: Based on the intent recognition results and service chain node information, relevant documents and information fragments are retrieved from the power knowledge base.

[0084] Ranking stage: The search results are ranked based on indicators such as semantic similarity, timeliness, and relevance to select the most relevant information.

[0085] Generation phase: Using a finely tuned large language model from the power sector, the retrieved information is integrated with standardized dialogue templates to generate natural and fluent responses.

[0086] Optimization phase: Based on customer feedback and historical dialogue data, continuously optimize the script templates and generation strategies.

[0087] The script generation process employs the following methods: 1) Emotion-aware script adjustment: By analyzing the emotional characteristics (such as speech rate, tone, and keywords) in customer conversations in real time, the tone and expression of the script are dynamically adjusted. For example, when a customer is detected to be agitated, a more concise and reassuring response is automatically switched. 2) Multi-turn dialogue script management: A coherent multi-turn dialogue flow can be designed to ensure the completeness of information collection and the fluency of the dialogue. 3) Script caching optimization: Distributed caching technology is used to store high-frequency script templates, and the priority is dynamically adjusted according to the access frequency. For example, scripts for high-frequency questions such as "How to pay the electricity bill" are cached in memory, reducing the response time to milliseconds.

[0088] The script optimization module supports the following functions: 1) A / B testing: Test multiple versions of the script with the same intent and select the version with the highest conversion rate and satisfaction. 2) Real-time quality inspection: Automatically evaluate script quality through an intelligent quality inspection system, identify inaccurate or non-compliant answers, and trigger manual review. 3) Dynamic learning: Continuously update the script library and optimize the generation strategy based on customer feedback and dialogue results to achieve self-evolution of the scripts.

[0089] The business processor integrates with the power business system through interfaces and data processing modules. During the intent processing unit's processing, if interaction with the power business system (such as the customer service business platform) is required, the business processor is responsible for sending the processing request to the customer service business platform and receiving and processing the returned results.

[0090] The business processing and execution layer connects to the power service business platform, achieving a closed loop from customer intent recognition to response to business processing requirements. This includes: 1) Interface adaptation layer: adapting to the business system interfaces of different power companies to achieve standardized calls. 2) Process engine: automatically triggering or assisting manual completion of business processes based on service chain node information, such as querying electricity bills, dispatching work orders, and generating power supply contracts. 3) Status monitoring: real-time monitoring of business processing status and feeding the results back to the intent processing unit, supporting dynamic adjustment of scripts and guidance strategies. 4) Security control: based on power business security standards, setting up multi-level verification and auditing mechanisms for sensitive operations (such as modifying customer information and adjusting fees).

[0091] The key innovation of the business processor lies in closely integrating intent recognition results with standardized business processes, ensuring that the semantic understanding capabilities of the large language model are combined with the standardization of power business. For example, in the electricity bill inquiry scenario, the query for a customer's electricity bill information will be performed using the customer's account number as the primary identifier and returned to the intent processing unit; in the fault reporting scenario, fault information will be recorded and a work order will be issued.

[0092] The knowledge base is a database used to store information such as power business knowledge, frequently asked questions, policies, and regulations. It provides knowledge support for the intent processing unit and business processor. When handling customer issues, relevant information can be retrieved from the knowledge base to help generate accurate solutions.

[0093] The intent service chain library is a chain composed of multiple intent processing units and related data flows. It describes the complete process of handling customer intents, ensuring that customer requests are processed comprehensively and accurately. Each intent service chain consists of multiple types of nodes, including query, judgment, information processing, data retrieval, and response types. Taking the "electricity bill inquiry" service chain as an example, its directed graph structure includes nodes for obtaining the customer number and querying the electricity bill by account number, as well as multiple branch paths formed based on whether the user provides an account number. In the directed graph, node jump control is implemented through the weight and direction of the edges. For example, when the user cannot provide an account number, according to the path selection mechanism of the directed graph, the user is automatically redirected to the node that queries the phone number or ID number. This path selection mechanism, combined with dual-labeling, can dynamically calculate the optimal path based on context variables and execution conditions.

[0094] The service chain library uses structured data representation. Each node contains the following information: • Node type: such as intent confirmation, inquiry, judgment, information processing, data retrieval, and response. • Node conditions: the context state and information required to trigger the node. • Node operation: the specific operation performed (such as calling a business interface or generating a script). • Node script instructions: standardized script templates and alternative scripts. • Node branches: possible exceptions and their handling branches.

[0095] The construction of the intent service chain library can follow power industry standards to ensure the compliance and executability of processes. The library also includes power business priority definitions, such as fault reporting being a high-priority business (response time within seconds), while electricity bill inquiries are a medium-priority business (response time within seconds), helping the system to allocate resources rationally.

[0096] After identifying the customer's intent, the matching service chain is loaded. The context is initialized, and the service chain is started. Processing begins step by step from the first node: invoking the script according to the node type, collecting responses, updating variables, and judging conditions. In each round of dialogue, the system decides whether to jump to the next node or revert to a previous question based on the customer's response. After all necessary information has been collected, the system executes the necessary interface calls to obtain the results. The final response script is output to inform the customer that the request has been completed. After all nodes have been completed, the session is closed or the customer is prompted whether to continue with other business.

[0097] The customer service business platform includes a core processing unit for customer service operations, covering functions such as customer information inquiry, electricity bill inquiry, and business processing, and interfaces with the electricity marketing system. It provides the business processor with actual business data and logic support to complete specific electricity business operations.

[0098] The module for transferring requests to human customer service includes judgment criteria and a human customer service interface. During intent recognition or processing, if the system determines that it cannot accurately handle the customer's problem or the customer proactively requests it, it will transfer requests that meet the conditions for transferring to a human customer service representative, ensuring that the customer's problem can be properly resolved. This disclosure can be widely applied to multiple customer service terminals such as power company customer service centers, the 95598 hotline, the State Grid App, enterprise WeChat official accounts, and virtual intelligent customer service robots, enabling intelligent processing of business scenarios such as electricity bill inquiries, fault reporting, account transfer processing, payment reminders, and intelligent electricity usage suggestions.

[0099] The operation of the intelligent customer service system for electricity mainly includes the following steps: 1) Customer input: Customers express their service needs through voice, text or video.

[0100] 2) Intent recognition: The system analyzes customer input, identifies the initial intent, and extracts key information.

[0101] 3) Service chain matching: Select the most matching standardized process from the service chain library based on the identified intent.

[0102] 4) Response script generation: Based on service chain node information and customer history dialogue, dynamically generate response scripts that conform to power business specifications.

[0103] 5) Service Information Collection: Guide customers to provide necessary information through multiple rounds of dialogue to ensure the completeness of data required for business processing.

[0104] 6) Business Execution: Connect to the power business platform to automatically or manually complete business operations.

[0105] 7) Result Feedback: Based on the standard instructions in the service chain, the large language model generates personalized results for the customer based on the business processing results and provides feedback to the customer, thus ending the dialogue.

[0106] 8) Continuous optimization: Collect dialogue data and customer feedback to continuously optimize the intent recognition model and service chain.

[0107] The system also includes the following key components: 1) Identity verification: Employing power industry-specific verification methods such as telephone verification, facial recognition for electricity application, and single-certificate application to ensure the authenticity and validity of customer identities. 2) Anomaly handling: When customer information is missing or intent is ambiguous, the system automatically switches to an anomaly handling branch (such as human customer service), or ensures the service process continues through multiple rounds of guidance or emotional reassurance. 3) Dynamic adjustment: Based on the dialogue status and changes in customer emotions, the system dynamically adjusts its scripts and guidance strategies; for example, automatically shortening the number of dialogue rounds when a customer shows impatience.

[0108] Customer intent recognition first preprocesses the customer input, including speech recognition, text cleaning, and word segmentation, to ensure the input data meets the requirements of subsequent processing. Then, a large language model is used to analyze the customer input and extract preliminary intent and key information (slots).

[0109] Service Intent Verification Module: This module matches and verifies the recognition results from the large language model against the service intent of the power knowledge base, correcting any potential recognition errors. The power knowledge base service contains information such as power business terminology, process specifications, and policies and regulations, ensuring that intent recognition complies with power supply service standards.

[0110] A "service chain" is a service operation link built around a specific customer intent. It is a sequence of predefined task nodes (such as question guidance, information collection, interface calls, and response generation) with clear process control logic and script instruction templates to drive the system to complete business processing. The service chain library is a collection of all service chains, grouped and indexed by intent for easy calling, maintenance, and expansion.

[0111] Each service chain consists of the following elements: 1) Intent Identifier: The customer intent corresponding to this service chain, such as "query electricity bill". 2) Processing Node Chain: An ordered array of task nodes, each node executing a service step. 3) Node Type: Divided into inquiry, judgment, information processing, data retrieval, and response types. 4) Node Phrases: The standard question or answer requirements issued by the node, indicating the response requirements of the large language model. 5) Execution Condition: A Boolean expression used to determine whether to execute the node. 6) Input or Output Parameters: Used to store intermediate variables such as customer account number, name, etc. 7) Interface Call Configuration: If the node needs to call external services, configure the interface path, parameter mapping, and other information. 8) Node Jump Control: Set the next node ID or branch node.

[0112] The design of the intent service chain library follows these principles: Comprehensive Business Coverage: Covering major power business scenarios such as electricity bill inquiries, fault reporting, business expansion applications, and complaints / suggestions. Standardized Processes: Each service chain corresponds to a standardized business process, ensuring seamless integration with the power company's internal business systems. Robust Anomaly Handling: Each service chain includes normal processes and anomaly handling branches, such as strategies for handling authentication failures and missing information. Dynamic Extensibility: Supporting dynamic adjustment and optimization of business processes through knowledge graph updates and service chain editing tools.

[0113] For example, the smart electricity customer service method of this disclosure embodiment is illustrated using an electricity bill inquiry intent service chain. Taking "electricity bill inquiry intent" as an example, the service chain design is as follows: Intent name: Electricity bill inquiry; Node type: Response type; Node operation: Executed by calling the service sub-chain. Service chain: Obtain the time period for the customer to inquire about electricity bills -> Obtain the account number -> Inquire about the electricity bill for the specified time using the account number. Node dialogue instruction: The account number you are inquiring about is $$1, and the electricity bill for the time $$2 is $$3? Node return parameter: Intent response reply text string.

[0114] Service Subchain: Service Subchain Name: Get the time period for customer electricity bill inquiry; Node Type: Information Processing; Node Operation: Indicate the date range in the large language model dialogue; Node Dialogue Instruction: Do you want to inquire about the electricity bill for $$ month? Node Branch 1: Obtain the date the customer needs to inquire about from the dialogue. Node Branch 2: If there is no result for Branch 1, retrieve the month of the current default date from the system as the date range. Node Return Parameter: The time period for querying electricity bill.

[0115] Service Subchain Name: Get Account Number. Node Type: Inquiry-based; Node Operation: Indicates the date range in the large language model dialogue. Node Input Parameters: Interactive dialogue. Node Phrases: What is your phone number / address / ID number, and what is the corresponding account number? Node Branch 1: Confirm by matching the account number with the incoming call; Node Branch 2: Inquire about the customer's account number and confirm; Node Branch 3: Inquire about the phone number, automatically retrieve the account number and confirm the account holder's name; Node Branch 4: If there is no account number or phone number, retrieve the account number and confirm the account holder's name using the ID number; Node Branch 5: Inquire about the electricity address, retrieve and confirm the account number. Node Return Parameter: Account Number.

[0116] Account Number Inquiry: Service Sub-chain Name: Account Number Inquiry Electricity Bill. Node Type: Data Retrieval Class; Node Operation: Call the account number inquiry electricity bill interface. Node Input Parameters: Account Number, Time Period; Node Dialogue Command: Do you want to inquire about the electricity bill for $$ month? Node Branch 1: Returns the electricity bill number; Node Branch 2: Returns an empty value; Node Return Parameter: Electricity Bill.

[0117] For example, in the case of a customer querying their electricity bill, in the customer interaction layer, the customer queries the information sent on May 27, 2025: "I want to check my electricity bill for last month."

[0118] Step 1: Text input or voice is captured in the {customer interaction layer} and passed to the subject intent guidance and recognition module. This module combines the customer context and the latest interaction history to recognize the subject intent through a large language model. Step 2: The intent is successfully identified as "Electricity Bill Inquiry". The intent is sent to the intent service chain scheduling center. The intent service chain scheduler schedules the intent processing unit according to the resource situation. The intent processing unit matches the intent service chain as "Electricity Bill Inquiry". According to the intent "Electricity Bill Inquiry", the service chain requires 3 service sub-chains: get the time period of the customer's electricity bill inquiry -> get the account number -> query the electricity bill for the specified time through the account number.

[0119] Step 3: Initiate the "Check Electricity Bill" intent service sub-chain: 1) Obtaining the time period sub-chain for customer electricity bill inquiry: Using the intent guidance generation module, based on the instruction to "obtain the time period for customer electricity bill inquiry," the module defaults to the current time and generates the guidance message, "Do you want to inquire about your electricity bill for April 2025?". Step 3: The guidance message is fed back to the customer through the customer interaction layer for confirmation: "Do you want to inquire about your electricity bill for April 2025?" If the customer answers "Yes," the process proceeds to the [Electricity Bill Inquiry] intent service chain (obtaining the account number). Otherwise, the inquiry service sub-chain is used to obtain the time period information.

[0120] 2) Obtaining the account number service sub-chain processing: The service chain steps are as follows: (1) Branch 1: If it is a telephone call, the account number matching interface of the power service business platform can be called from the caller's binding information to automatically match the account number, receive the return value, and initiate account name confirmation: Guiding words: "The account name you registered for this number is Zhang San. Is this you?" (2) Branch 2: Service chain two: If the caller does not match the account number, guide the customer to enter the account number: Guiding words: "Can you provide the account number on the electricity bill?" (3) Branch 3: If there is no account number, guide the customer to provide the phone number and then match the account number: Guiding words: "Please provide the mobile phone number you used when registering for the electricity service, and we will match the account number for you." (4) Branch 4: If the customer does not have an account number or phone number, they can ask for identity information or service address: Guiding words: "You can provide your ID card number or electricity address, and we will help you find it." 3) Querying electricity charges for a specified time by account number: The service sub-chain steps are as follows: Based on the obtained time period and customer account number, call the electricity charge query interface of the power customer service business platform, input the query time period and account number, and return the customer's electricity consumption in May 2025: electricity consumption: 180 kWh, electricity charge: 96.3 yuan; The charge query node's script instruction is passed to the large language model, which processes the intent return value: Intent response return value: "Your account number 123456789 consumed a total of 180 kWh of electricity in April 2025, with a total electricity charge of 96.3 yuan." Step 5: The intent service chain scheduling center provides the intent response return value to the customer interaction layer and feeds it back to the customer, completing the intent response. Step 6: The customer interaction layer records the interaction history to the customer interaction history database. As can be seen, the method of this disclosure embodiment is highly controllable, can avoid the illusion caused by the free generation of large language models, and provides standardized and traceable service processes; it provides precise services by combining intent recognition and service chain engines to make services more in line with customers' actual needs; it has an efficient closed loop by using the configuration method of the service chain to drive the large language model to automatically advance the process, supplement information, and complete closed-loop business processing; and it can be continuously optimized by supporting the use of customer interaction history records to continuously optimize service levels.

[0121] The method in this disclosure utilizes natural language processing technology and a knowledge graph in the power sector to perform semantic analysis on the text input by the customer, identify the customer's current intent, and categorize it into a preset power customer service intent system. Based on the identified customer intent, the corresponding service process link is matched from a pre-built standardized service chain library. Each service chain contains multiple standard service nodes, defining standard service actions and response instruction logic for that intent. Combining the current node status of the service chain and the historical dialogue context, the large language model generates a dynamic dialogue response string that conforms to business specifications according to the service chain instructions, ensuring that the response content is both natural and fluent and consistent with business logic. Through interface integration with the power business platform, relevant business data (such as electricity bills, electricity usage records, repair work orders, etc.) is extracted in real time according to the customer intent and service chain node requirements, guiding the large language model to generate relevant dialogue content to form the final response. After the service is completed, service process data is automatically recorded, and machine learning algorithms are used to continuously optimize the intent recognition model and service chain matching strategy, improving the system's intelligence level and service quality.

[0122] Through the above-mentioned technical solutions, this invention realizes the intelligentization, standardization, and personalization of the power customer service process, significantly improving service response efficiency and customer satisfaction, while reducing the illusion risk of large language models in power scenarios and enhancing the security and reliability of services.

[0123] As can be seen, in contrast, this disclosure proposes a power intelligent customer service method based on a power big language model. By establishing an intent service chain, realizing multi-turn dialogue and contextual understanding, accurately identifying customer intent and matching standardized service chains, it effectively solves the above-mentioned problems existing in the prior art.

[0124] It should be noted that the method of this embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this embodiment, and the multiple devices will work together to generate video to complete the method described.

[0125] It should be noted that the above description describes some embodiments of this disclosure. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0126] Based on the same technical concept, corresponding to any of the above embodiments, this disclosure also provides an intelligent power customer service device, see [link to relevant documentation]. Figure 5 The intelligent power customer service device includes: The acquisition module is used to acquire first user input information, which includes at least one of voice data, text data, or video data. The semantic analysis module is used to perform semantic analysis on the first user input information to determine the first user service information; the first user service information is used to indicate service tasks. The process matching module is used to match the corresponding service operation process based on the first user service information. The service operation process includes a directed graph formed by multiple service operation nodes. The process execution module is used to select a corresponding path along the directed graph based on the node input information corresponding to the service operation node in order to execute the service task; the node input information is obtained by semantic analysis based on the second user input information of the user for the service operation node; The result return module is used to return the execution result of the service task to the user.

[0127] For ease of description, the above apparatus is described in terms of its functions, divided into various modules. Of course, in implementing this disclosure, the functions of each module can be implemented in one or more software and / or hardware.

[0128] The apparatus described above is used to implement the corresponding smart electricity customer service method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0129] Based on the same technical concept, corresponding to the methods of any of the above embodiments, this disclosure also provides a non-transitory computer-readable storage medium that stores computer instructions for causing the computer to execute the intelligent power customer service method as described in any of the above embodiments.

[0130] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0131] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the smart power customer service method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0132] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this disclosure (including the claims) is limited to these examples; within the framework of this disclosure, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this disclosure as described above, which are not provided in detail for the sake of brevity.

[0133] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this disclosure, the provided drawings may or may not show well-known power / ground connections to integrated circuit (IC) chips and other components. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this disclosure, and this also takes into account the fact that the details of implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this disclosure will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this disclosure, it will be apparent to those skilled in the art that the embodiments of this disclosure can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0134] Although this disclosure has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0135] This disclosure is intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A smart electricity customer service method, comprising: Obtain first user input information, wherein the first user input information includes at least one of voice data, text data, or video data; Semantic analysis is performed on the first user input information to determine the first user service information; the first user service information is used to indicate service tasks. Based on the first user service information, a corresponding service operation process is matched, and the service operation process includes a directed graph formed by multiple service operation nodes; Based on the node input information corresponding to the service operation node, a corresponding path is selected along the directed graph to execute the service task; The node input information is obtained through semantic analysis based on the second user input information of the user for the service operation node. The execution result of the service task will be returned to the user.

2. The method according to claim 1, wherein, The node input information is obtained through semantic analysis based on the second user input information for the service operation node, including: The node service information corresponding to the service operation node is output to the user; the node service information is used to prompt the user for the information required to perform the service operation of the service operation node. Receive the second user input information from the user regarding the node service information; Semantic analysis is performed on the second user input information to obtain the node input information.

3. The method according to claim 1, wherein, Based on the node input information corresponding to the service operation node, a corresponding path is selected along the directed graph, including: In response to detecting that the node input information satisfies the first condition of the service operation node, a first path associated with the service operation node in the directed graph is selected; In response to detecting that the node input information satisfies the second condition of the service operation node, a second path associated with the service operation node in the directed graph is selected; In response to the detection that the node input information does not meet any of the conditions of the service operation node, a service anomaly is determined.

4. The method according to claim 1, further comprising: The corresponding service processor is called based on the service operation interface corresponding to the service operation node to process the node input information and execute the service operation corresponding to the service operation node.

5. The method according to claim 1, wherein, Semantic analysis is performed on the first user input information to determine the first user service information, including: The first user input information is processed to obtain intermediate input information; Semantic features are extracted based on the intermediate input information to obtain intermediate input features; Preliminary user service information is determined based on the intermediate input features; The preliminary user service information is verified based on the preset service information; Upon successful verification of the preliminary user service information, the preliminary user service information is determined to be the first user service information. If the verification of the preliminary user service information fails, the preliminary user service information is updated based on the preset service information to obtain the first user service information.

6. The method according to claim 2, wherein, Output the node service information corresponding to the service operation node to the user, including: Determine the user status based on the first user input information and / or the second user input information; In response to the user's state not meeting the state conditions, update the content and / or the number of times the node service information is updated; And / or, The node service information is generated based on the user's historical interaction information, the context information of the current interaction, and the node information template corresponding to the service operation node.

7. The method of claim 1, further comprising at least one of the following: In response to the inability to determine the first user service information based on the first user input information, guidance information is generated to guide the user to input supplementary information; Based on the supplementary user input information and the first user input information, semantic analysis is performed to determine the first user service information; In response to the inability to determine the first user service information based on the supplementary user input information and the first user input information, a service anomaly is determined. Based on the exception type and service exception policy of the service exception, the corresponding exception handling operation is executed.

8. A smart electricity customer service device, comprising: The acquisition module is used to acquire first user input information, which includes at least one of voice data, text data, or video data. The semantic analysis module is used to perform semantic analysis on the first user input information to determine the first user service information; the first user service information is used to indicate service tasks. The process matching module is used to match the corresponding service operation process based on the first user service information. The service operation process includes a directed graph formed by multiple service operation nodes. The process execution module is used to select the corresponding path along the directed graph based on the node input information corresponding to the service operation node in order to execute the service task; The node input information is obtained through semantic analysis based on the second user input information of the user for the service operation node. The result return module is used to return the execution result of the service task to the user.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as claimed in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method of any one of claims 1 to 7.

Images