Knowledge question-answering method based on large model, electronic equipment and storage medium

By using a combination of knowledge model and large model in the dialogue system, users can obtain user input and historical dialogue status information, which solves problems that are not professional and accurate in answering electrical equipment questions, and achieves a more accurate and safe dialogue process, which improves users Experience.

CN119025642APending Publication Date: 2024-11-26QINGDAO HAIER TECH +2

Patent Information

Application Number
CN202411048688.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-01
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

The existing dialogue system based on large-scale pre-trained language models. When answering electrical equipment-related questions, the knowledge base coverage is incomplete and may contain error information, resulting in inadequate answers and accurate answers, and the 'illusion' phenomenon of the big model may pass errors. Information affects user trust and device security.

Method used

Using a knowledge question-and-answer method based on the big model, by obtaining user input and historical dialogue status information, using the knowledge model to determine the answer data, including knowledge data and control instructions, ensuring the accuracy and security of the answer, using the big model to generate answer information, and Adjust the device status if necessary.

Benefits of technology

It improves the accuracy and coherence of the conversation, enhances the user experience, ensures the professionalism and security of the answers, and reduces the risk of user operation errors.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025642A_ABST
    Figure CN119025642A_ABST
Patent Text Reader

Abstract

The invention discloses a knowledge question answering method based on a large model, electronic equipment and a storage medium, and relates to the technical field of smart home / smart home. The knowledge question and answer method based on the large model comprises the steps that question information and historical dialogue state information input by a user are obtained, and the historical dialogue state information comprises question information and answer information of historical dialogue rounds; based on the question information and the historical dialogue state information, a knowledge model is adopted to determine answer data corresponding to the question information, and the answer data comprises knowledge data and / or a control instruction corresponding to the question information; and based on the answer data, using the large model to generate answer information, and feeding back the answer information to the user. According to the method, the knowledge model and the large model are combined to deeply understand the question information and then make professional answers, so that the accuracy and fluency of the knowledge question-answering process are ensured.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of smart home / intelligent family technology, and more specifically, to a knowledge question-answering method, electronic device, and storage medium based on a large model. Background Art

[0002] When building a dialogue system using large-scale pre-trained language models (such as GPT), although these models have demonstrated excellent language understanding and generation capabilities through training with massive data, can interact with users naturally and fluently, and reduce dependence on human intervention, they still face many challenges in practical applications. For example, the limitations of the knowledge base. In the external knowledge base or internal memory mechanism that the current large language model relies on, the professional knowledge about electrical equipment is often not fully covered and may contain erroneous or outdated information. This results in the system being unable to provide accurate and reliable information when answering users' specific questions about electrical equipment, thereby affecting user satisfaction and trust. In addition, the "hallucination" phenomenon of the large model itself, that is, when the model generates answers, it sometimes outputs information that is inconsistent with facts or logic, that is, "fabricating" content to fill its knowledge gaps or adapt to the context. In the dialogue system, the impact of the hallucination phenomenon is particularly significant. It may not only mislead users and convey wrong information, but is also more likely to cause users to operate based on these wrong answers, thereby causing damage to electrical equipment. Summary of the invention

[0003] In order to overcome the above-mentioned defects, the present application is proposed to provide a large-model-based knowledge question-answering method, electronic device and storage medium that solves or at least partially solves the technical problem that the knowledge question-answering system involving electrical appliance knowledge is not professional and accurate enough when answering user questions.

[0004] In a first aspect, the present application provides a knowledge question-answering method based on a big model, the method comprising: obtaining question information and historical dialogue status information input by a user, the historical dialogue status information comprising question information and answer information of historical dialogue rounds; based on the question information and the historical dialogue status information, using a knowledge model to determine answer data corresponding to the question information, wherein the answer data includes knowledge data and / or control instructions corresponding to the question information; based on the answer data, using a big model to generate answer information, and feeding back the answer information to the user.

[0005] In a technical solution of the above-mentioned knowledge question and answer method based on a large model, the knowledge model is used to determine the answer data corresponding to the question information based on the question information and the historical dialogue state information, including: based on the question information and the historical dialogue state information, obtaining the knowledge data corresponding to the question information in the knowledge model; based on the knowledge data, determining the answer data corresponding to the question information, wherein the answer data contains the knowledge data corresponding to the question information.

[0006] In a technical solution of the above-mentioned knowledge question and answer method based on a large model, the knowledge model is used to determine the answer data corresponding to the question information based on the question information and the historical dialogue state information, including: based on the question information and the historical dialogue state information, obtaining the knowledge data corresponding to the question information in the knowledge model; based on the knowledge data, judging whether to obtain the device status corresponding to the question information; if so, obtaining the device status corresponding to the question information; based on the knowledge data and the device status, determining the answer data corresponding to the question information, the answer data containing the knowledge data and / or control instructions corresponding to the question information.

[0007] In a technical solution of the above-mentioned knowledge question and answer method based on a big model, the knowledge data corresponding to the question information in the knowledge model is obtained based on the question information and the historical dialogue state information, including: based on the question information and the historical dialogue state information, using a big model to perform semantic understanding on the question information to determine the intention corresponding to the question information; based on the intention corresponding to the question information, the knowledge data corresponding to the question information in the knowledge model is obtained.

[0008] In a technical solution of the above-mentioned knowledge question and answer method based on a big model, the big model is used to generate answer information based on the answer data and feedback it to the user, including: when the answer data contains knowledge data and control instructions corresponding to the question information, adjusting the device state corresponding to the question information based on the control instructions; based on the knowledge data and the control instructions, the big model is used to generate answer information and feedback it to the user.

[0009] In a technical solution of the above-mentioned knowledge question and answer method based on a large model, before using a knowledge model to determine the answer data corresponding to the question information based on the question information and the historical dialogue state information, the method also includes: using a pre-trained classification model to determine whether the question information belongs to the field of knowledge question and answer; if it does, then based on the question information and the historical dialogue state information, using a knowledge model to determine the answer data corresponding to the question information.

[0010] In a technical solution of the above-mentioned knowledge question and answer method based on a big model, after the answer information is fed back to the user, the method also includes: updating the historical dialogue state information based on the question information and the answer information of the current dialogue round; obtaining the question information re-entered by the user; based on the updated dialogue state information and the re-entered question information, obtaining the answer data corresponding to the re-entered question information, forming multiple consecutive rounds of dialogue until the big model generates answer information that meets preset conditions.

[0011] In a technical solution of the above-mentioned knowledge question and answer method based on a large model, the method also includes: obtaining at least one of the instruction manual data, explanation video data, operation video data, fault question and answer manual data, service manual data, and standard operating procedure data related to the electrical equipment; based on at least one of the instruction manual data, explanation video data, operation video data, fault question and answer manual data, service manual data, and standard operating procedure data related to the electrical equipment, creating a knowledge model, the knowledge model including multiple knowledge branches, each knowledge branch including multiple nodes.

[0012] In a second aspect, an electronic device is provided, comprising at least one processor and at least one memory, wherein the memory is suitable for storing a plurality of program codes, and the program codes are suitable for being loaded and run by the processor to execute the knowledge question answering method based on a big model as described in any one of the technical solutions of the above-mentioned knowledge question answering method based on a big model.

[0013] In a third aspect, a computer-readable storage medium is provided, which stores a plurality of program codes, wherein the program codes are suitable for being loaded and run by a processor to execute the knowledge question answering method based on a big model as described in any one of the technical solutions of the above-mentioned knowledge question answering method based on a big model.

[0014] The above one or more technical solutions of the present application have at least one or more of the following beneficial effects:

[0015] The present application discloses a knowledge question-answering method based on a big model, which specifically includes: obtaining question information and historical dialogue status information input by a user, wherein the historical dialogue status information includes question information and answer information of historical dialogue rounds; based on the question information and the historical dialogue status information, using a knowledge model to determine answer data corresponding to the question information, wherein the answer data includes knowledge data and / or control instructions corresponding to the question information; based on the answer data, using a big model to generate answer information, and feeding the answer information back to the user.

[0016] By recording historical conversation status information, this application can accurately grasp the context of the user's questions, thereby gaining a deeper understanding of the user's intentions and needs. The knowledge model is used to determine the answer data corresponding to the question information, which solves the illusion problem of the large model and ensures the accuracy of the generated answer data. The large model is used to generate answer information based on the answer data, making the conversation process more natural and smooth, which not only enhances the coherence of the conversation, but also improves the overall user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0019] Figure 1 This is a flowchart of the main steps of the knowledge question answering method based on a large model according to an embodiment of the present application;

[0020] Figure 2 It is a schematic diagram of a question-answering process of a knowledge question-answering method based on a large model according to an embodiment of the present application;

[0021] Figure 3 It is a schematic diagram of the interactive process of a knowledge question answering method based on a large model according to an embodiment of the present application;

[0022] Figure 4 It is a schematic diagram of the main structural block diagram of an electronic device according to an embodiment of the present application. List of reference numerals:

[0023] 11: processor; 12: memory. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0025] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0026] Existing dialogue systems built on large-scale pre-trained language models (such as GPT) have many problems. For example, the limitations of the knowledge base. The external knowledge base or internal memory mechanism that the current large models rely on often does not fully cover the professional knowledge about electrical equipment and may contain erroneous or outdated information. This results in the system being unable to provide accurate and reliable information when answering users' specific questions about electrical equipment, thus affecting user satisfaction and trust. In addition, the "hallucination" phenomenon of the large model itself, that is, when the model generates answers, it sometimes outputs information that is inconsistent with facts or logic, that is, "fabricating" content to fill its knowledge gaps or adapt to the context. This may cause users to operate based on these wrong answers, thereby causing damage to electrical equipment.

[0027] To this end, the present application provides a knowledge question and answer method based on a big model, which specifically includes: obtaining question information and historical dialogue status information input by the user, the historical dialogue status information including question information and answer information of historical dialogue rounds; based on the question information and the historical dialogue status information, using a knowledge model to determine answer data corresponding to the question information, wherein the answer data includes knowledge data and / or control instructions corresponding to the question information; based on the answer data, using a big model to generate answer information, and feeding the answer information back to the user.

[0028] By recording and updating historical conversation status information, this application can accurately grasp the context of the user's questions, thereby more deeply understanding the user's intentions and needs. The knowledge model is used to determine the answer data corresponding to the question information, which solves the illusion problem of the large model and ensures the accuracy of the generated answer data. Using the large model to generate answer information based on the answer data makes the conversation process more natural and smooth, which not only enhances the coherence of the conversation, but also improves the overall user experience.

[0029] See attached Figure 1 , Figure 1FIG. 1 is a flow chart of the main steps of the knowledge question answering method based on a large model according to an embodiment of the present application. Figure 1 As shown, a knowledge question answering method based on a large model of the present application mainly includes the following steps S101 to S103.

[0030] Step S101: obtaining question information input by a user and historical dialogue state information, wherein the historical dialogue state information includes question information and answer information of historical dialogue rounds.

[0031] In this embodiment, the pre-trained big model receives the question information input by the user in the current dialogue round, and the question information can be the question information input by the user in various ways and converted into text form. The historical dialogue state information includes the dialogue records of the historical dialogue rounds between the user and the big model before the current dialogue round.

[0032] Step S102: Based on the question information and the historical dialogue state information, a knowledge model is used to determine answer data corresponding to the question information, wherein the answer data includes knowledge data and / or control instructions corresponding to the question information.

[0033] In this embodiment, answer data is obtained based on the question information, historical dialogue state information, and the knowledge model. The answer data includes electrical appliance knowledge data and control instructions related to the question information obtained from the knowledge model.

[0034] Step S103: Based on the answer data, use the big model to generate answer information, and feed back the answer information to the user.

[0035] Based on the above steps S101-S103, this application uses a knowledge model to determine the answer data corresponding to the question information based on the question information and historical dialogue state information obtained in the current dialogue round, and can accurately grasp the context of the user's question through the historical dialogue state information, so as to more deeply understand the user's intentions and needs. The knowledge model is used to determine the answer data corresponding to the question information, which solves the illusion problem of the large model and ensures the accuracy of the generated answer data. Using the large model to generate answer information based on the answer data and communicate with the user through the answer information makes the dialogue process more natural and smooth, which not only enhances the coherence of the dialogue, but also improves the overall user experience.

[0036] The above steps S101 to S103 are further explained below.

[0037] With respect to step S101, question information input by the user and historical dialogue state information are obtained, wherein the historical dialogue state information includes question information and answer information of historical dialogue rounds.

[0038] In this embodiment, the pre-trained big model receives the question information input by the user in the current dialogue round. The question information can be question information input by the user in various ways and converted into text form, such as keyboard input, voice input, etc. Question information such as "Why does the water purifier not work?", "How to turn on the air conditioner?", etc. The historical dialogue state information includes the dialogue records of the historical dialogue rounds between the user and the big model before the current dialogue round.

[0039] With respect to step S102, based on the question information and the historical dialogue state information, a knowledge model is used to determine answer data corresponding to the question information, wherein the answer data includes knowledge data and / or control instructions corresponding to the question information.

[0040] In some embodiments, based on the question information and the historical conversation state information, a knowledge model is used to determine answer data corresponding to the question information, including: based on the question information and the historical conversation state information, obtaining knowledge data corresponding to the question information in the knowledge model; based on the knowledge data corresponding to the question information, determining answer data corresponding to the question information, wherein the answer data includes the knowledge data corresponding to the question information.

[0041] The knowledge model is a structured knowledge base built based on knowledge data related to electrical equipment. The knowledge model includes the contents of multiple knowledge branches, each knowledge branch contains multiple nodes, and the knowledge branches and nodes are determined based on the knowledge data related to electrical equipment. The knowledge data related to electrical equipment can be data such as user guides or troubleshooting procedures. When communicating with users, users can be guided based on the knowledge branches and nodes in the knowledge model, and multiple rounds of dialogue interaction with users can be achieved based on accurate knowledge data.

[0042] Specifically, based on the question information, the knowledge branch matching the question information in the knowledge model is queried, and the content of which node on the knowledge branch is output as knowledge data in the current dialogue round in combination with the historical dialogue state information. The knowledge data can be in various forms such as text, pictures, and videos. The obtained knowledge data is used as the answer data corresponding to the question information.

[0043] For example, when the user inputs the question "Why does the under-sink water purifier not work?", the knowledge model searches for relevant knowledge data based on the question information. First, it is identified that there are three types of under-sink water purifiers, namely, type A water purifiers, type B water purifiers, and type C water purifiers.

[0044] Since each type of water purifier may have its own specific cause of failure and solution, further dialogue with the user is needed to determine which type of under-sink water purifier the user is asking about. Therefore, the acquired knowledge data - that is, the information that there are three types of under-sink water purifiers - is first fed back as preliminary answer data to guide the user to determine the specific type of water purifier, ensuring that subsequent answers are more accurate and effective.

[0045] In some embodiments, based on the question information and the historical conversation status information, a knowledge model is used to determine answer data corresponding to the question information, including: based on the question information and the historical conversation status information, obtaining knowledge data corresponding to the question information in the knowledge model; based on the knowledge data, determining whether to obtain a device status corresponding to the question information; if so, obtaining the device status corresponding to the question information; based on the knowledge data and the device status, determining answer data corresponding to the question information, the answer data including knowledge data and / or control instructions corresponding to the question information.

[0046] Specifically, based on the question information and historical dialogue status information, the queried knowledge data may need to refer to the device status information of the electrical equipment related to the question information. Therefore, it is determined based on the knowledge data whether it is necessary to obtain the device status corresponding to the question information. If so, the device status required to answer the user's question is obtained through a preset data interface. Then, based on the knowledge data and the device status, the answer data corresponding to the question information is determined. The answer data may include knowledge data, control instructions, or knowledge data and control instructions.

[0047] For example, when the question information input by the user is "Why can't the washing machine door be opened?", the knowledge data related to the washing machine door failure is searched in the knowledge model based on the question information. The knowledge model includes multiple knowledge branches related to the washing machine door failure, for example, checking the child lock status and checking the washing machine water level. These are first output as knowledge data, and the need to obtain the device status of the washing machine is determined based on the checking of the child lock status and checking the water level of the washing machine in the knowledge data, and then the device status of the washing machine is obtained.

[0048] If the child lock status in the washing machine's device status is on, the washing machine door is prohibited from opening when the child lock is on, and answer data including control instructions and knowledge data is obtained based on the knowledge data and the device status. The control instruction is to switch the child lock status to the off state, and the knowledge data is "the cause of the washing machine door failure is that the child lock is turned on."

[0049] If the water level of the washing machine exceeds a preset threshold in the device status of the washing machine, and the door remains locked when there is water in the washing machine, the answer data determined based on the knowledge data only includes the knowledge data, that is, "the cause of the washing machine door failure is water in the washing machine."

[0050] If the water level of the washing machine in the device state of the washing machine does not exceed the preset threshold, and the child lock state is off, the answer data determined based on the knowledge data only includes the control instruction for controlling the opening of the washing machine door.

[0051] In some embodiments, the obtaining of knowledge data corresponding to the question information in the knowledge model based on the question information and the historical conversation state information includes: based on the question information and the historical conversation state information, using a large model to perform semantic understanding on the question information to determine the intention corresponding to the question information; based on the intention corresponding to the question information, obtaining the knowledge data corresponding to the question information in the knowledge model.

[0052] Specifically, before acquiring knowledge data, we first use the big model to perform semantic analysis on the question information of the current dialogue round in combination with the historical dialogue state information to determine the specific intention of the user's question. Then, based on the determined intention corresponding to the question information, we acquire the knowledge data corresponding to the question information in the knowledge model.

[0053] With respect to step S103, answer information is generated based on the answer data using a large model, and the answer information is fed back to the user.

[0054] In some embodiments, based on the answer data, answer information is generated using a large model and fed back to the user, including: when the answer data contains knowledge data and control instructions corresponding to the question information, adjusting the device state corresponding to the question information based on the control instructions; based on the knowledge data and the control instructions, answer information is generated using a large model and fed back to the user.

[0055] Specifically, when the output answer data includes knowledge data and control instructions, the device state of the electrical device corresponding to the question information is adjusted according to the control instructions, and based on the knowledge data and control instructions, a large model is used to generate answer information and feedback to the user.

[0056] For example, when the question information input by the user is "Why can't the washing machine door be opened?", and the child lock status in the obtained device status of the washing machine is on, the answer data determined based on the question information includes knowledge data and control instructions. The control instruction is to switch the child lock status to off, and the knowledge data is "The cause of the failure of the washing machine door is that the child lock is turned on."

[0057] Adjust the device status of the washing machine based on the control instruction and switch the child lock status of the washing machine to the off state. The large model generates an answer message based on the knowledge data and the control instruction. The answer message will explain to the user the possible reasons why the washing machine door cannot be opened and indicate that the relevant fault has been resolved. For example, the answer message can be "The reason why your washing machine door cannot be opened is because the child lock state is turned on. Now, we have switched the child lock state to the off state for you. Please try to open the washing machine door again."

[0058] In some embodiments, when the answer data contains knowledge data and control instructions corresponding to the question information, the large model is used to generate answer information based on the knowledge data and control instructions and feedback to the user. The answer information will explain to the user the possible reasons why the washing machine door cannot be opened, and prompt the solution to the fault, and confirm with the user whether it is necessary to help the user solve the relevant fault. Then, according to the user's further instructions, the device status of the washing machine is adjusted based on the control instructions.

[0059] In some embodiments, before using a knowledge model to determine answer data corresponding to the question information based on the question information and the historical conversation state information, the method also includes: using a pre-trained classification model to determine whether the question information belongs to the field of knowledge question and answer; if it does, then based on the question information and the historical conversation state information, using a knowledge model to determine answer data corresponding to the question information.

[0060] Specifically, before determining the answer data corresponding to the question information, first determine whether the question information input by the user belongs to the field of knowledge question and answer based on the pre-trained classification model. For example, the question information input by the user is "Why does the washing machine not perform single dehydration?", "Why does the light of the range hood not light up?", etc. This type of information is information in the fault question and answer manual data and belongs to the field of knowledge question and answer. The question information input by the user is "Turn on the light" or "Play music". This type of control information does not belong to the field of knowledge question and answer. If it is determined that the question information belongs to the field of knowledge question and answer, the knowledge model is used to determine the answer data corresponding to the question information based on the question information and the historical dialogue state information.

[0061] In some embodiments, after feeding back the answer information to the user, the method further includes: updating the historical dialogue state information based on the question information and the answer information of the current dialogue round; obtaining the question information re-entered by the user; based on the updated dialogue state information and the re-entered question information, obtaining the answer data corresponding to the re-entered question information, to form multiple consecutive rounds of dialogue until the large model generates answer information that meets preset conditions.

[0062] Specifically, after each output of answer information, the historical dialogue state information is updated based on the question information and answer information of the current dialogue round, so that when the user's question information of the next dialogue round is obtained, the user's intention of asking questions can be better understood based on the historical dialogue state information. And the current state of the dialogue process is determined based on the historical dialogue state information to output answer data that conforms to the current dialogue process.

[0063] The question information input again by the user is obtained, and semantic understanding of the question information input again is performed based on the updated dialog state information to determine the intention of the user to ask the question again.

[0064] Based on the user's intention to ask the question again and the updated dialogue status information, the answer data corresponding to the re-entered question information is obtained in the knowledge model, and answer information is generated based on the answer data and fed back to the user, thus forming multiple rounds of continuous dialogue with the user until the answer information output by the large model meets the preset conditions, and the dialogue ends.

[0065] Whether the output answer information meets the preset conditions can be judged based on the knowledge data contained in the answer information. For example, when the knowledge data contained in the answer information generated by the large model points to the final node of a knowledge branch in the knowledge model, it means that the knowledge of the knowledge branch has been fully provided or the question has been fully answered, and the judgment dialogue can end.

[0066] Alternatively, it is determined whether the user has specific statements or behaviors during the conversation (such as not responding for a long time, entering commands such as "exit" or "stop") to indicate the intention to end the conversation. When these statements or behaviors are detected, it is determined that the conversation can be ended.

[0067] Alternatively, it is determined whether the user directly gives an instruction to end the conversation during the conversation, such as "Thank you, I understand" or "OK, goodbye". When it is detected that the user has given an instruction similar to the desire to end the conversation, it is determined that the conversation can be ended.

[0068] See attached Figure 2 , Figure 2 It is a schematic diagram of the question-answering process of a knowledge question-answering method based on a large model according to an embodiment of the present application.

[0069] like Figure 2 As shown, the user inputs the question information "What is the normal desalination rate of the under-sink water purifier?".

[0070] The large model obtains knowledge data corresponding to the question information in the knowledge model through question information and historical status dialogue information. First, it recognizes that under-sink water purifiers are divided into three knowledge branches based on type, namely Type A water purifiers, Type B water purifiers and Type C water purifiers. Since each type of water purifier may have its own specific standards, further dialogue with the user is required to determine which type of under-sink water purifier the user is asking about.

[0071] Therefore, the acquired knowledge data, namely the information that there are three types of under-sink water purifiers, is first sent to the big model as preliminary answer data. The big model generates answer information based on the answer data, "Is your water purifier type A, type B, or type C?", to guide the user to determine the specific type of water purifier, ensuring that subsequent answers are more accurate and effective. The historical dialogue status information is updated based on the question information and answer information of the current dialogue round.

[0072] While waiting for the answer information to be fed back, the large model can provide the user with a status prompt message, for example, "Querying for you, please wait."

[0073] The user inputs the question information "A type water purifier" again.

[0074] Based on the question information and historical status dialogue information, the big model obtains the knowledge data of the relevant nodes for determining the desalination rate in the knowledge branch corresponding to "Type A water purifier" in the knowledge model, and obtains "the desalination rate is calculated based on the total dissolved solids value of the raw water and the total dissolved solids value of the purified water", and sends the knowledge data as answer data to the big model.

[0075] The large model generates answer information based on the answer data, "What are the total dissolved solids (TDS) values ​​of raw water and purified water?", and guides the user to enter the total dissolved solids values ​​of raw water and purified water required to calculate the desalination rate. The historical dialogue status information is updated based on the question information and answer information of the current dialogue round.

[0076] The user inputs the question information again: "The price of raw water is 200, and the price of purified water is 20."

[0077] Based on the question information and historical status dialogue information, the big model obtains the calculation method of the desalination rate in the knowledge branch corresponding to "Type A water purifier" in the knowledge model, and calculates the desalination rate to be 90%. Then, based on 90%, it judges whether the desalination rate is within the normal range, and obtains the knowledge data "The desalination rate is 90%, normal range", and sends the knowledge data as answer data to the big model.

[0078] The large model generates answer information based on the answer data, "The desalination rate of type A water purifier is 90%, which is within the normal range." And based on the question information and answer information of the current dialogue round, the historical dialogue state information is updated.

[0079] In some embodiments, the method also includes: obtaining at least one of instruction manual data, explanation video data, operation video data, troubleshooting question and answer manual data, service manual data, and standard operating procedure data related to the electrical equipment; creating a knowledge model based on at least one of the instruction manual data, explanation video data, operation video data, troubleshooting question and answer manual data, service manual data, and standard operating procedure data related to the electrical equipment, the knowledge model including multiple knowledge branches, and each knowledge branch including multiple nodes.

[0080] Specifically, the knowledge model is a structured knowledge base built based on a variety of data resources related to electrical equipment. These data resources include but are not limited to instruction manual data, explanation video data, operation video data, troubleshooting question and answer manual data, service manual data, and standard operating procedure data.

[0081] When building the model, the data is first cleaned and preprocessed to remove redundant and erroneous information. Then, key knowledge points and structured information are extracted from the data, and then organized into multiple knowledge branches and nodes, each of which represents a specific knowledge point, fault handling process node or operation step, etc.

[0082] When communicating with users, the knowledge model serves as a guiding tool. Based on the user's question information, relevant knowledge data and control instructions are retrieved from the knowledge branches and nodes of the knowledge model. Knowledge data can be in various forms such as text, pictures, and videos.

[0083] When interacting with users in multiple rounds of dialogue, the knowledge model can dynamically adjust the path and method of knowledge retrieval based on the user's question information and historical dialogue status information. For example, if the user asks about a washing machine failure in the first round of dialogue, the knowledge model can retrieve the corresponding knowledge data from the knowledge branch related to the failure. If the user has questions about this knowledge data or needs further help, the knowledge model can guide the user to a more detailed knowledge branch or node and provide more in-depth information and solutions.

[0084] See attached Figure 3 , Figure 3 It is a schematic diagram of the interactive process of a knowledge question-answering method based on a large model according to an embodiment of the present application.

[0085] like Figure 3As shown, the user inputs question information to the large model through text input or voice input, and the question information can be any question in the field of electrical appliance knowledge question answering.

[0086] After receiving the question information, the large model processes the question information based on the stored historical dialogue state information and the knowledge model, and obtains the knowledge data and / or control instructions related to the question information in the knowledge model. Specifically, the knowledge data related to the question information in the knowledge model is obtained, the device status associated with the question information is obtained based on the knowledge data, and the answer data corresponding to the question information is determined based on the knowledge data and the device status, and the answer data includes the knowledge data and / or control instructions corresponding to the question information.

[0087] The big model generates answer information based on the answer data output by the knowledge model and feeds it back to the user, and then updates the historical dialogue status information based on the answer information and question information.

[0088] It is understood by those skilled in the art that all or part of the processes in the method for implementing any of the above embodiments of the present application can also be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of each of the above method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable storage medium may include: any entity or device, medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory, random access memory, electric carrier signal, telecommunication signal and software distribution medium, etc. that can carry the computer program code. It should be noted that the content contained in the computer-readable storage medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable storage media do not include electric carrier signals and telecommunication signals.

[0089] Furthermore, the present application also provides an electronic device.

[0090] See attached Figure 4 , Figure 4 1 is a main structural block diagram of an electronic device according to an embodiment of the present application. Figure 4As shown, in an electronic device embodiment according to the present application, the electronic device includes at least one processor 11 and at least one memory 12, the memory 12 can be configured to store a program for executing the knowledge question answering method based on a large model of the above method embodiment, and the processor 11 can be configured to execute the program in the memory, which includes but is not limited to the program for executing the knowledge question answering method based on a large model of the above method embodiment. For ease of explanation, only the parts related to the embodiment of the present application are shown. For specific technical details not disclosed, please refer to the method part of the embodiment of the present application. The electronic device can be an electronic device formed by various electronic devices.

[0091] Furthermore, the present application also provides a computer-readable storage medium. In a computer-readable storage medium embodiment according to the present application, the computer-readable storage medium can be configured to store a program for executing the large-model-based knowledge question-answering method of the above-mentioned method embodiment, and the program can be loaded and run by the processor to implement the above-mentioned large-model-based knowledge question-answering method. For ease of explanation, only the parts related to the embodiment of the present application are shown. For specific technical details not disclosed, please refer to the method part of the embodiment of the present application. The computer-readable storage medium can be a storage device formed by various electronic devices. Optionally, the computer-readable storage medium in the embodiment of the present application is a non-temporary computer-readable storage medium.

[0092] Further, it should be understood that since the setting of each module is only for illustrating the functional units of the device of the present application, the physical devices corresponding to these modules may be the processor itself, or a part of the software in the processor, a part of the hardware, or a part of the combination of software and hardware. Therefore, the number of each module in the figure is only schematic.

[0093] Further, it should be understood that the method of the present application is widely used in smart home (Smart Home), smart home, smart home device ecology, smart home (Intelligence House) ecology and other whole-house intelligent digital control application scenarios. The above-mentioned network may include but is not limited to at least one of the following: wired network, wireless network. The above-mentioned wired network may include but is not limited to at least one of the following: wide area network, metropolitan area network, local area network, and the above-mentioned wireless network may include but is not limited to at least one of the following: WIFI (Wireless Fidelity, Wireless Fidelity), Bluetooth. Terminal devices may not be limited to PCs, mobile phones, tablet computers, smart air conditioners, smart range hoods, smart refrigerators, smart ovens, smart stoves, smart washing machines, smart water heaters, smart washing equipment, smart dishwashers, smart projection equipment, smart TVs, smart clothes drying racks, smart curtains, smart audio and video, smart sockets, smart speakers, smart speakers, smart fresh air equipment, smart kitchen and bathroom equipment, smart bathroom equipment, smart sweeping robots, smart window cleaning robots, smart mopping robots, smart air purification equipment, smart steamers, smart microwave ovens, smart kitchen treasures, smart purifiers, smart water dispensers, smart door locks, etc.

[0094] It is understood by those skilled in the art that each module in the device can be adaptively split or merged. Such splitting or merging of specific modules will not cause the technical solution to deviate from the principle of the present application, and therefore, the technical solutions after splitting or merging will fall within the protection scope of the present application.

[0095] So far, the technical solutions of the present application have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present application is obviously not limited to these specific embodiments. Without departing from the principles of the present application, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present application.

Claims

1. A knowledge question answering method based on a large model, characterized in that: The method comprises: Acquire the question information and historical dialogue status information input by the user, wherein the historical dialogue status information includes the question information and answer information of the historical dialogue rounds; Based on the question information and the historical dialogue state information, using a knowledge model to determine answer data corresponding to the question information, wherein the answer data includes knowledge data and / or control instructions corresponding to the question information; Based on the answer data, answer information is generated using a large model, and the answer information is fed back to the user.

2. The knowledge question answering method based on a large model according to claim 1 is characterized in that: The step of using a knowledge model to determine answer data corresponding to the question information based on the question information and the historical dialogue state information includes: Based on the question information and the historical dialogue state information, acquiring knowledge data corresponding to the question information in the knowledge model; Based on the knowledge data, answer data corresponding to the question information is determined, and the answer data includes the knowledge data corresponding to the question information.

3. The knowledge question answering method based on a large model according to claim 1, characterized in that: The step of using a knowledge model to determine answer data corresponding to the question information based on the question information and the historical dialogue state information includes: Based on the question information and the historical dialogue state information, acquiring knowledge data corresponding to the question information in the knowledge model; Based on the knowledge data, determining whether to obtain the device status corresponding to the question information; If yes, then obtaining the device status corresponding to the question information; Based on the knowledge data and the device status, answer data corresponding to the question information is determined, and the answer data includes the knowledge data and / or control instructions corresponding to the question information.

4. The knowledge question answering method based on a large model according to claim 2 or 3, characterized in that: The acquiring, based on the question information and the historical dialogue state information, knowledge data corresponding to the question information in the knowledge model includes: Based on the question information and the historical dialogue state information, a large model is used to perform semantic understanding on the question information to determine the intent corresponding to the question information; Based on the intention corresponding to the question information, knowledge data corresponding to the question information in the knowledge model is acquired.

5. The knowledge question answering method based on a large model according to claim 3 is characterized in that: Based on the answer data, the large model is used to generate answer information and feed it back to the user, including: In a case where the answer data includes knowledge data and control instructions corresponding to the question information, adjusting the device state corresponding to the question information based on the control instructions; Based on the knowledge data and the control instructions, a large model is used to generate answer information and feed it back to the user.

6. The knowledge question answering method based on a large model according to claim 1, characterized in that: Before determining answer data corresponding to the question information by using a knowledge model based on the question information and the historical dialogue state information, the method further includes: Using a pre-trained classification model, determining whether the question information belongs to the field of knowledge question answering; If yes, then based on the question information and the historical dialogue state information, a knowledge model is used to determine answer data corresponding to the question information.

7. The knowledge question answering method based on a large model according to claim 1, characterized in that: After feeding back the answer information to the user, the method further includes: Update the historical dialogue state information based on the question information and the answer information of the current dialogue round; Get the question information entered by the user again; Based on the updated dialogue state information and the re-input question information, answer data corresponding to the re-input question information is obtained to form multiple consecutive rounds of dialogue until the large model generates answer information that meets preset conditions.

8. The knowledge question answering method based on a large model according to claim 1, characterized in that: The method further comprises: Obtain at least one of instruction manual data, explanation video data, operation video data, troubleshooting question and answer manual data, service manual data, and standard operating procedure data related to the electrical equipment; A knowledge model is created based on at least one of the instruction manual data, explanation video data, operation video data, troubleshooting question and answer manual data, service manual data, and standard operating procedure data related to the electrical equipment. The knowledge model includes multiple knowledge branches, and each knowledge branch includes multiple nodes.

9. An electronic device comprising at least one memory and at least one processor, characterized in that: The memory stores a computer program, and the processor is configured to execute the large model-based knowledge question answering method according to any one of claims 1 to 8 through the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein the program, when running, executes the large model-based knowledge question answering method according to any one of claims 1 to 8.

Citation Information

Patent Citations

  • Question-answer matching method based on historical dialogue information and related device

    CN111858854A

  • Multi-round dialogue method and device based on artificial intelligence and storage medium

    CN113806508A

  • Electronic equipment, interactive knowledge acquisition method of electronic equipment and interactive system

    CN117807196A

  • Washing machine control method and device, washing machine and storage medium

    CN118029090A

  • Household equipment control method and device, electronic equipment and storage medium

    CN118034415A

Cited By

  • Home security system based on multi-mode perception and semantic linkage control

    CN120595624A

  • Home intelligent question and answer and design method and device based on large model and medium

    CN121051850A

  • A large model-based home intelligent question answering and design method, device and medium

    CN121051850B