Intelligent heat supply customer service question answering method and system based on large model and knowledge base

Abstract

The invention belongs to the technical field of heat supply customer service, and provides an intelligent heat supply customer service question answering method and system based on a large model and a knowledge base, and the method comprises the steps: obtaining an intelligent heat supply knowledge base; receiving a user question, identifying the received user question based on the large model, and extracting key elements of the user question; and performing semantic retrieval in the obtained intelligent heat supply knowledge base according to the extracted key elements, obtaining answers to the user questions by adopting retrieval enhancement, and completing intelligent heat supply customer service questions and answers based on the large model and the knowledge base.

Description

Technical Field

[0001] This invention belongs to the field of heating customer service technology, specifically relating to a smart heating customer service question-and-answer method and system based on a large model and knowledge base. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Urban centralized heating systems mainly consist of three parts: heat sources, heating networks, and heat users. They utilize heat sources such as thermal power plants and district boiler rooms, and supply heat to users through heating pipelines and other facilities. With continuous social development, accelerated urbanization, and the gradual improvement of people's living standards, the scale of urban heating pipelines has become increasingly large, which in turn increases the difficulty of managing various aspects of the heating system.

[0004] The overall level of intelligent management and control in grassroots heating enterprises is relatively low, with a predominantly extensive management model and cumbersome existing customer service management systems. Existing smart heating customer service systems suffer from the following shortcomings: reliance on human agents or standardized FAQs makes it impossible to address diverse user inquiries, resulting in low response efficiency; limited customer service knowledge coverage makes it difficult to integrate professional materials such as equipment manuals, operating procedures, and historical work orders; inability to provide differentiated and personalized services based on different user roles (such as residents and maintenance engineers); and lagging knowledge updates, making it difficult to promptly reflect operational data and newly emerging fault cases. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a smart heating customer service question-and-answer method and system based on a large model and knowledge base. In a smart heating scenario, by utilizing a large model and knowledge enhancement methods, and by constructing a dynamically updated heating knowledge base, introducing a role recognition module, and implementing a retrieval enhancement generation mechanism, the system achieves structured parsing of natural language questions and multi-source knowledge invocation, thereby improving the accuracy of responses and the system response rate, while reducing the workload of human customer service representatives.

[0006] According to some embodiments, the first solution of the present invention provides a smart heating customer service question-and-answer method based on a large model and knowledge base, which adopts the following technical solution: A smart heating customer service question-and-answer method based on a large model and knowledge base includes: Access the smart heating knowledge base; Receive user questions, identify the received user questions based on a large model, and extract the key elements of the user questions; Based on the extracted key elements, semantic retrieval is performed in the acquired smart heating knowledge base. Search enhancement is used to obtain answers to user questions, thus completing smart heating customer service Q&A based on a large model and knowledge base.

[0007] As a further technical limitation, a smart heating customer service Q&A method based on a large model and knowledge base also includes obtaining differentiated answers to user questions according to the user's role type, collecting user feedback information on the obtained differentiated user question answers, feeding the collected user feedback information into the acquired smart heating knowledge base, and iteratively optimizing the acquired smart heating knowledge base; wherein, the user's role type includes at least residents and maintenance personnel.

[0008] As a further technical limitation, in the process of acquiring the smart heating knowledge base, heating system text information including at least documents, operation logs and fault work orders is collected. The collected text information is vectorized to obtain a vector database. A dynamic knowledge base is generated based on the obtained vector database, thus obtaining the smart heating knowledge base.

[0009] As a further technical limitation, in the process of identifying the received user questions based on a large model, the large model is used to identify and parse user intent. Specifically, the received user questions are preprocessed, and the user role type is determined based on the preprocessed user questions. User intent is then identified and parsed for different user role types. In the process of identifying and parsing different user role types, the preprocessed user questions are sequentially subjected to coarse classification, slot filling, ambiguity resolution, and quality gating to obtain the user intent.

[0010] As a further technical limitation, the semantic retrieval includes at least document segmentation and indexing, candidate recall, business filtering, reordering, context packing, and security trimming.

[0011] As a further technical limitation, the process of obtaining answers to user questions includes at least prompt templates, tool function calls, controlled generation, role-based templates, traceable output, and write-back annotations.

[0012] According to some embodiments, the second aspect of the present invention provides a smart heating customer service Q&A system based on a large model and knowledge base, employing the following technical solution: A smart heating customer service Q&A system based on a large model and knowledge base includes: The acquisition module is configured to acquire the smart heating knowledge base; The extraction module is configured to receive user questions, identify the received user questions based on a large model, and extract the key elements of the user questions. The question-and-answer module is configured to perform semantic retrieval in the acquired smart heating knowledge base based on the extracted key elements, and obtain answers to user questions using retrieval enhancement, thus completing smart heating customer service question-and-answer based on a large model and knowledge base.

[0013] According to some embodiments, a third aspect of the present invention provides a computer-readable storage medium, employing the following technical solution: A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the intelligent heating customer service Q&A method based on a large model and knowledge base as described in the first aspect of the present invention.

[0014] According to some embodiments, the fourth aspect of the present invention provides an electronic device, which adopts the following technical solution: An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor. When the processor executes the program, it implements the steps in the intelligent heating customer service Q&A method based on a large model and knowledge base as described in the first aspect of the present invention.

[0015] According to some embodiments, the fifth aspect of the present invention provides a computer program product, which adopts the following technical solution: A computer program product includes software code, wherein the program in the software code performs the steps of the intelligent heating customer service Q&A method based on a large model and knowledge base as described in the first aspect of the present invention.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: Because traditional human agents or rule-based FAQ systems cannot efficiently respond to diverse user questions, this invention combines a large model with a knowledge base to achieve rapid understanding and automatic response to resident inquiries, thereby improving response efficiency. Existing customer service systems often can only answer a limited number of fixed questions, making it difficult to cover professional knowledge such as equipment manuals, operating specifications, and historical work orders. To solve the problem of insufficient knowledge coverage, this invention constructs a dynamically updated industry knowledge base and performs semantic enhancement retrieval with a large model to ensure comprehensive customer service knowledge coverage.

[0017] Traditional customer service cannot provide differentiated responses based on different user identities (residents or maintenance personnel). To solve the problem of lack of personalized service, this invention can generate targeted personalized response content through user intent recognition and role classification.

[0018] Heating operation parameters and maintenance cases are constantly changing. Manual maintenance of the knowledge base is costly and not updated in a timely manner. To solve the problem of lagging knowledge updates, this invention supports real-time updates of the knowledge base by automatically collecting operation data and work order logs, ensuring that the answers are consistent with the latest situation. Attached Figure Description

[0019] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0020] Figure 1 This is a flowchart of the intelligent heating customer service question-and-answer method based on a large model and knowledge base in Embodiment 1 of the present invention. Figure 2 This is a structural block diagram of the intelligent heating customer service Q&A system based on a large model and knowledge base in Embodiment 2 of the present invention. Detailed Implementation

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0022] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0023] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0024] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0025] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.

[0026] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0027] Example 1 Embodiment 1 of this invention introduces a smart heating customer service question-and-answer method based on a large model and knowledge base.

[0028] like Figure 1 The method shown is a smart heating customer service question-and-answer approach based on a large model and knowledge base, including: Access the smart heating knowledge base; Receive user questions, identify the received user questions based on a large model, and extract the key elements of the user questions; Based on the extracted key elements, semantic retrieval is performed in the acquired smart heating knowledge base. Search enhancement is used to obtain answers to user questions, thus completing smart heating customer service Q&A based on a large model and knowledge base.

[0029] This embodiment, in a smart heating scenario, utilizes a large-scale model and knowledge enhancement methods to achieve efficient, accurate, and personalized customer service Q&A, significantly improving service efficiency and user experience while reducing the workload of human customer service representatives. Specifically, it combines a large-scale model with a dynamic knowledge base to provide intelligent Q&A methods that enable rapid, accurate, and multi-round natural language responses to heating user inquiries; it achieves structured management and efficient retrieval of multi-source heterogeneous data (documents, logs, work orders, operating parameters) within the heating industry, expanding the scope of customer service knowledge coverage; it constructs a personalized response mechanism based on user role recognition, providing usage guidance to residents and fault diagnosis and repair suggestions to maintenance personnel; and it establishes a real-time knowledge base update mechanism, automatically integrating operational data and new cases to improve the system's timeliness and adaptability.

[0030] This embodiment combines industry-specific knowledge structures for dynamic updates, that is, unifying multi-source heterogeneous data such as equipment manuals, operation manuals, heat exchange station / secondary network operation logs, maintenance work orders, historical room temperature and payment data into a "heating field entity (equipment—site—circuit—parameter—operating condition—fault—work order—measure)" and adding incremental data to the database in real time (minute level), instead of static FAQs; By combining user role-aware response orchestration, the same intent is addressed with different granularities for residents and maintenance engineers: residents provide "operable steps + safety tips," while engineers provide "step-by-step location + quantified thresholds + tool list + verification points." This is a deep industry differentiation from general property customer service / online agents. The RAG2.0 retrieval chain is adopted, which is a hybrid retrieval of dense vector retrieval (embedding) and sparse retrieval (BM25). Then, a cross encoder is used to rearrange the data and apply constraints of "site / device / time window / data freshness" before finally sending it to the large model for generation. The generation end requires mandatory reference to source fragments and timestamps to reduce illusions and improve auditability.

[0031] A closed-loop learning approach is adopted, which involves writing back "user feedback / handling results / follow-up work orders / changes in measured parameters" to form verifiable samples. These samples are used for retrieval weight labeling, template optimization, and domain-specific small sample fine-tuning, forming an engineered closed loop of "knowledge-question-answering-verification-increment".

[0032] It should be noted that this embodiment addresses the customer service needs of smart heating systems by combining resident-side issues, maintenance work orders, equipment manuals, and operational data to form an industry-specific knowledge base. Compared with traditional general customer service or property management customer service, it emphasizes the integration of heating operation conditions, thermodynamic parameters, and pipeline / heat exchange station data.

[0033] As one or more implementation methods, this embodiment collects heating system text information including at least document data, operation logs, and fault work orders, performs vectorization processing on the collected text information to obtain a vector database, and generates a dynamic knowledge base based on the obtained vector database, thus obtaining a smart heating knowledge base.

[0034] It should be noted that this embodiment adopts a dynamic knowledge base update mechanism, that is, by collecting operation logs and work order data in real time, the knowledge base is automatically updated to ensure that the knowledge is synchronized with the system status; this further overcomes the shortcomings of existing technologies that mostly rely on "static knowledge bases" or "manual maintenance" and lack automated update capabilities.

[0035] To achieve dynamic knowledge base updates, this embodiment constructs a closed-loop mechanism of "data collection—knowledge update—intelligent question answering—user feedback—knowledge iteration." Specifically, after obtaining answers to user questions, differentiated answers are derived based on the user's role type. User feedback on these differentiated answers is collected, and this feedback is fed back into the acquired smart heating knowledge base for iterative optimization. The user role types include at least residents and maintenance personnel.

[0036] As one or more implementation methods, in this embodiment, the user inputs a natural language question via WeChat / Mini Program; then, a large model is used to perform intent recognition on the input text, extracting the question type (e.g., payment, equipment, abnormal room temperature) and key elements (e.g., location, equipment number); specifically: (1) Preprocessing: noise reduction, sentence segmentation, regular expression extraction of numbers (household number, table number, station), place name, and time; ASR converts speech to text.

[0037] (2) Role determination: Multimodal determination is made based on the source (resident mini-program / engineer app / enterprise WeChat), account binding and text features (determination score ≥ τ_r is effective).

[0038] (3) Coarse classification: Lightweight classifier (such as intent label: payment / temperature control / room temperature not up to standard / leak / valve abnormal / water pump abnormal / heat source side abnormal / consultation...) Softmax to get p(intent|x).

[0039] (4) Slot filling: Named entity recognition + rule fusion extracts stations, equipment (valves / pumps / heat exchangers / heat source units), parameters (supply and return water temperature, pressure difference, opening degree), time windows, etc.

[0040] (5) Ambiguity resolution: If key information is missing or conflicting (confidence level < τ_i), raise 1-2 clarification questions (such as "Please confirm the name of the heat exchange station / time period / room temperature measurement point").

[0041] (6) Quality gate control: The overall score s = α·p(intent) + β·p(role) + γ·p(slots completeness rate); if s < τ_s, then proceed to manual or minimum clarification.

[0042] This embodiment identifies received user questions based on a large model and extracts key elements of the user questions. The user intent recognition module preprocesses the input text, including word segmentation, noise reduction, named entity recognition, and slot extraction; then, it uses an intent classification model to determine the category of the question and outputs intent labels and key entities; the system further determines the user role (resident or maintenance personnel) based on the input channel, user history behavior, and semantic features, and sets a confidence threshold. When the recognition confidence is lower than the preset value, a clarification prompt or transfer to a human agent is triggered to ensure recognition accuracy.

[0043] As one or more implementation methods, this embodiment performs semantic retrieval in the smart heating knowledge base based on the recognition results to obtain a set of candidate answers; then, the candidate answers and user input are input into a large model, and the final response is generated using the retrieval-enhanced generation (RAG) method; that is, given the intent and slots, the most relevant and up-to-date evidence paragraphs are retrieved from massive documents / logs for the large model to generate the answer; specifically: (1) Document segmentation and indexing: Instructions / rules / work orders / logs are segmented into semantic paragraphs (300–800 words) and a dual index is generated: dense vector (domain embedding) + sparse inverted index (BM25).

[0044] (2) Candidate Recall (K1): Use query=f(intent,slots,data summary of the most recent T hours of the site) to search two paths at the same time, and take the top-K1 (e.g., 30).

[0045] (3) Business filtering: Filter by site / device / time window / document type; prioritize near real-time information for operational information; prioritize historical work orders with the same type / symptom.

[0046] (4) Rearrangement (K2): The cross-encoder calculates the relevance of the candidates; the features of "freshness / source credibility / whether it contains operation steps / whether it contains threshold" are added and linearly fused and rearranged, and the top-K2 is taken (e.g., 6).

[0047] (5) Context Packet: Control the total context length, retain the source ID + timestamp of each segment, and insert the site’s recent time series statistics (average / variance of room temperature in the last 24 hours, supply and return water temperature difference, pressure difference).

[0048] (6) Security trimming: Remove fields containing personal information / privacy data (such as mobile phone numbers) and desensitize them.

[0049] This embodiment employs a Retrieval Augmentation (RAG) mechanism to achieve knowledge retrieval and answer generation. First, textual information such as heating system-related documents, operation logs, and maintenance work orders is semantically segmented and vectorized, establishing both a dense vector index and a sparse inverted index. Upon receiving a user's question, the system performs a hybrid recall based on the question's semantic features within the dual index database, and then reorders the recall results using a cross-encoder. During reordering, document freshness, source credibility, and time window weights are introduced to prioritize the latest and most relevant knowledge fragments. Before generation, the system performs security filtering on the search results, removing sensitive fields and selecting the final candidate set based on confidence thresholds. The large model combines these knowledge fragments to generate the answer, and the output is annotated with the information source number and timestamp to ensure content traceability and verification.

[0050] As one or more implementation methods, the large model in this embodiment generates natural language responses based on candidate knowledge and contextual information; if the user is a resident, it outputs operation instructions or payment reminders; if the user is an operations and maintenance personnel, it outputs fault cause analysis and solution steps; provided there is evidence, it outputs executable, traceable, and verifiable answers, and distinguishes between resident / operations and maintenance personnel presentations; specifically: (1) Prompt template: Role commands (resident / engineer); Task instructions (first provide the diagnostic conclusion, then the steps and basis, and finally the risks and review). Evidence constraints (only contexts are allowed, and the source number and time must be indicated); Output format (JSON / Markdown sections, including "reason - steps - required tools - judgment threshold - review points - when to transfer to manual review").

[0051] (2) Tool / function call: Call the site’s near time series characteristic function before generation and add the recent load / temperature difference / pressure difference anomaly summary.

[0052] (3) Controlled generation: Temperature ≤ 0.3, opening must be standardized with terminology (valve opening, supply and return water temperature, pressure difference unit is unified); trigger a clarifier for missing slots.

[0053] (4) Characterization template: Resident version: Operation in 3 steps, with necessary safety tips and a repair reporting portal; Engineer version: Step-by-step location tree (e.g., "secondary network end - circulating pump - balancing valve - heat exchanger fouling"), each step includes judgment thresholds and expected phenomena.

[0054] (5) Traceable output: List “evidence fragments and timestamps (doc_id# paragraph)” at the end; generate confidence score (retrieval consistency + generation consistency).

[0055] (6) Write back and label: The engineer clicks "Valid / Invalid" and can generate a "new case" for the repair process with one click, which can be entered into the case library and vector library.

[0056] This embodiment combines data collection with knowledge base updates to achieve dynamic accumulation and updating of knowledge; it integrates the large model reasoning module with knowledge base retrieval to improve the accuracy and professionalism of answers; it uses an interactive interface to ensure smooth communication between users and the system; and it employs a feedback and optimization closed-loop learning mechanism to enable the system to continuously evolve.

[0057] It should be noted that this embodiment uses a dynamic knowledge base to incrementally maintain and update operation logs, maintenance work orders, and instruction documents; the RAG (Retrieval Enhanced Generation) module performs dense / sparse mixed retrieval and rearrangement from the knowledge base during the question-and-answer phase, outputting evidence fragments with source identifiers and timestamps; the role recognition module determines whether the user is a resident or maintenance personnel based on the input channel and semantic features, and selects the corresponding generation template and output format accordingly. These modules collaborate through a unified data interface to form a closed-loop process of "knowledge update—semantic retrieval—role-based generation" to ensure the traceability and consistency of the answers.

[0058] This embodiment utilizes a large-scale model combined with a knowledge base, allowing users to receive answers within 1-2 seconds of entering a question. Compared to the average response time of 30-60 seconds for human customer service, this significantly shortens the response cycle and improves efficiency. This embodiment integrates equipment manuals, operation manuals, historical work orders, and real-time data, expanding knowledge coverage to cover over 90% of common resident inquiries, while traditional FAQ systems typically cover less than 60%. This embodiment can identify user roles, providing residents with concise operation guidance and maintenance personnel with detailed fault diagnosis and handling steps, avoiding the generalization problem of the "question-and-answer" model and improving the targeted and personalized service capabilities. The knowledge base in this embodiment has a dynamic update mechanism, completing knowledge entry within 10 minutes of work order and operation log generation, significantly improving information timeliness compared to the weekly or monthly updates of traditional manual maintenance. In a typical heating company serving 100,000 households, this embodiment can replace more than 50% of the workload of human agents, and is expected to save approximately 20%-30%. This significantly reduces customer service operation costs and labor costs. Due to improved answer accuracy and faster response, the first-time resolution rate of user problems is expected to increase to over 85%, which is about 20 percentage points higher than the traditional method, greatly improving user satisfaction.

[0059] Example 2 Embodiment 2 of the present invention introduces a smart heating customer service Q&A system based on a large model and knowledge base.

[0060] like Figure 2 The system shown is a smart heating customer service Q&A system based on a large model and knowledge base, including: The acquisition module is configured to acquire the smart heating knowledge base; The extraction module is configured to receive user questions, identify the received user questions based on a large model, and extract the key elements of the user questions. The question-and-answer module is configured to perform semantic retrieval in the acquired smart heating knowledge base based on the extracted key elements, and obtain answers to user questions using retrieval enhancement, thus completing smart heating customer service question-and-answer based on a large model and knowledge base.

[0061] As one or more implementation methods, the intelligent heating customer service Q&A system based on a large model and knowledge base in this embodiment also includes: The inter-module connection and collaboration mechanism module is configured as follows: the acquisition module sends operation logs / work orders / documents to the knowledge base module for vectorization and dynamic updates; the question-answering module has a built-in RAG submodule that performs dense / sparse hybrid retrieval on the knowledge base and generates an evidence set after cross-sorting; the question-answering module is also connected to the role recognition submodule, selecting differentiated prompt templates and output structures based on the identified user roles; the feedback optimization module receives user confirmation and processing results and writes them back to the knowledge base module to update weights and indexes. These modules are connected via API / message queues, and the data flow is sequentially: "Collection → Vectorization / Update → Retrieval / Resorting → Generation → Feedback Writeback".

[0062] The detailed steps are the same as those of the intelligent heating customer service Q&A method based on a large model and knowledge base provided in Example 1, and will not be repeated here.

[0063] Example 3 Embodiment 3 of the present invention provides a computer-readable storage medium.

[0064] A computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the intelligent heating customer service Q&A method based on a large model and knowledge base as described in Embodiment 1 of the present invention.

[0065] The detailed steps are the same as those of the intelligent heating customer service Q&A method based on a large model and knowledge base provided in Example 1, and will not be repeated here.

[0066] Example 4 Embodiment 4 of the present invention provides an electronic device.

[0067] An electronic device includes a memory, a processor, and a program stored in the memory and running on the processor. When the processor executes the program, it implements the steps in the intelligent heating customer service Q&A method based on a large model and knowledge base as described in Embodiment 1 of the present invention.

[0068] The detailed steps are the same as those of the intelligent heating customer service Q&A method based on a large model and knowledge base provided in Example 1, and will not be repeated here.

[0069] Example 5 Embodiment 5 of the present invention provides a computer program product.

[0070] A computer program product includes software code, wherein the program in the software code executes the steps of the intelligent heating customer service Q&A method based on a large model and knowledge base as described in Embodiment 1 of the present invention.

[0071] The detailed steps are the same as those of the intelligent heating customer service Q&A method based on a large model and knowledge base provided in Example 1, and will not be repeated here.

[0072] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention can be implemented using various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0073] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0074] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0075] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0076] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0077] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

[0078] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A smart heating customer service question-and-answer method based on a large model and knowledge base, characterized in that, include: Access to the smart heating knowledge base; Receive user questions, identify the received user questions based on a large model, and extract the key elements of the user questions; Based on the extracted key elements, semantic retrieval is performed in the acquired smart heating knowledge base. Search enhancement is used to obtain answers to user questions, thus completing smart heating customer service Q&A based on a large model and knowledge base.

2. The intelligent heating customer service question-and-answer method based on a large model and knowledge base as described in claim 1, characterized in that, It also includes obtaining differentiated answers to user questions based on user role types, collecting user feedback on the differentiated user question answers, feeding the collected user feedback into the acquired smart heating knowledge base, and iteratively optimizing the acquired smart heating knowledge base; wherein, user role types include at least residents and operation and maintenance personnel.

3. The intelligent heating customer service question-and-answer method based on a large model and knowledge base as described in claim 1, characterized in that, In the process of acquiring the smart heating knowledge base, text information of the heating system, including at least documents, operation logs and fault work orders, is collected. The collected text information is then vectorized to obtain a vector database. A dynamic knowledge base is generated based on the obtained vector database, thus obtaining the smart heating knowledge base.

4. The intelligent heating customer service question-and-answer method based on a large model and knowledge base as described in claim 1, characterized in that, In the process of identifying the received user questions based on a large model, the large model is used to identify and parse user intent. Specifically, the received user questions are preprocessed, and the user role type is determined based on the preprocessed user questions. User intent is then identified and parsed for different user role types. In the process of identifying and parsing different user role types, the preprocessed user questions are sequentially subjected to coarse classification, slot filling, ambiguity resolution, and quality gating to obtain the user intent.

5. The intelligent heating customer service question-and-answer method based on a large model and knowledge base as described in claim 1, characterized in that, The semantic retrieval includes at least document segmentation and indexing, candidate recall, business filtering, reordering, contextual packaging, and security trimming.

6. The intelligent heating customer service question-and-answer method based on a large model and knowledge base as described in claim 1, characterized in that, The process of obtaining answers to user questions includes at least the following: prompt templates, utility function calls, controlled generation, role-based templates, traceable output, and write-back annotations.

7. A smart heating customer service Q&A system based on a large model and knowledge base, characterized in that, include: The acquisition module is configured to acquire the smart heating knowledge base; The extraction module is configured to receive user questions, identify the received user questions based on a large model, and extract the key elements of the user questions. The question-and-answer module is configured to perform semantic retrieval in the acquired smart heating knowledge base based on the extracted key elements, and obtain answers to user questions using retrieval enhancement, thus completing smart heating customer service Q&A based on a large model and knowledge base.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the intelligent heating customer service Q&A method based on a large model and knowledge base as described in any one of claims 1-6.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the program, it implements the steps of the intelligent heating customer service Q&A method based on a large model and knowledge base as described in any one of claims 1-6.

10. A computer program product, comprising software code, characterized in that, The program in the software code executes the steps of the intelligent heating customer service Q&A method based on a large model and knowledge base as described in any one of claims 1-6.

Images