Active question-answering system and method and refrigeration equipment

Abstract

The invention discloses an active question-answering system and method and refrigeration equipment, and belongs to the technical field of household appliances, the system comprises an interaction module, a control module, a question-answering reply module, a hardware detection module and a question recommendation module, the interaction module receives a first input and processes the first input to obtain a first text; the control module sends the first text to the question and answer reply module to determine a feedback text, and triggers the hardware detection module to collect sensing data to perform abnormal data detection to generate abnormal information; the question recommendation module generates a recommended question text based on the first text; the control module controls the interaction module to output a feedback text and abnormal information and / or a recommended question text; and the control module sends the first text to the question and answer reply module to determine a feedback text, triggers the hardware detection module and the question recommendation module to generate abnormal information and a recommended question text based on the first text, and finally controls the interaction module to output, so that abnormal early warning and active question asking are realized on the basis of normal feedback, and the user experience is improved.

Description

Technical Field

[0001] This application belongs to the field of household appliance technology, and in particular relates to an active question-and-answer system, method and refrigeration equipment. Background Technology

[0002] With the upgrading of refrigerators to intelligent features, products are constantly evolving towards anthropomorphism and personalization. Refrigerators are no longer limited to the single task of storage in daily family life. Following the continuous maturation of technologies such as artificial intelligence and big data, smart refrigerators are incorporating question-and-answer systems. After capturing user voice commands, smart refrigerators can quickly respond to them.

[0003] Smart question-and-answer systems for refrigerators enrich users' lives, but current systems suffer from the following user experience drawbacks: they rely on users' historical voice interaction information, making it impossible to recommend content when users first start using the refrigerator due to the lack of such historical data; alternatively, they rely on collecting users' historical life trajectories and pushing similar topics, which only allows for similar recommendations and cannot provide diverse and novel content; or they use hardware-based information collection for real-time monitoring and alerts, but if the hardware malfunctions, the question-and-answer function will be unavailable, and this reliance on hardware monitoring has limitations, potentially failing to cover all user question-and-answer scenarios and resulting in ineffective responses. Summary of the Invention

[0004] This application proposes an active question-answering system, method, and cooling device to solve the problem in related technologies where relying on users' historical voice interaction information, historical life trajectory, or hardware monitoring to achieve question-answering functions results in a poor user experience.

[0005] Firstly, this application provides an active question-answering system, comprising: an interaction module, a control module, a question-answering response module, a hardware detection module, and a question recommendation module, wherein...

[0006] The interaction module is used to receive a first input, process the first input, and obtain a first text;

[0007] The control module is used to send the first text to the question-and-answer response module to determine the feedback text, and to trigger the hardware detection module and the question recommendation module to run based on the first text;

[0008] The question-and-answer response module is used to determine the feedback text based on the first text;

[0009] The hardware detection module is used to collect hardware sensing data, perform abnormal data detection based on the hardware sensing data, generate abnormal information when abnormal data is detected, and send the abnormal information to the control module.

[0010] The question recommendation module is used to generate recommended question text based on the first text and send the recommended question text to the control module;

[0011] The control module is also used to control the interaction module to output the feedback text, as well as the abnormal information and / or recommended question text;

[0012] The interaction module is also used to output the feedback text, as well as the error information and / or recommended question text.

[0013] The proactive question-answering system provided in this application embodiment receives a first input through an interaction module, processes it to obtain a first text, and sends the first text to a question-answering response module to determine the feedback text. Based on the first text, the control module triggers the operation of a hardware detection module and a question recommendation module. The hardware detection module collects hardware sensing data to detect abnormal data. If abnormal data is detected, it generates abnormal information and sends it to the control module. The question recommendation module generates recommended text based on the first text and sends it to the control module. The control module controls the interaction module to output the feedback text, abnormal information, and / or recommended question text. By integrating the question recommendation module and the hardware detection module, the system achieves both normal feedback to the feedback text corresponding to the first input and abnormal warning and proactive question recommendation. It offers diverse question-answering modes, wide application scenarios, and enhances the user experience.

[0014] According to one embodiment of this application, the problem recommendation module includes:

[0015] The keyword extraction submodule is used to extract keywords from the first text to obtain a first keyword set.

[0016] The deep semantic derivation submodule is used to perform deep semantic derivation based on the first text to obtain the second text, wherein the second text includes deep semantic features;

[0017] The diffusion cross-domain submodule is used to generate text transfer features based on the first keyword set, the second text, and the user profile information corresponding to the first input;

[0018] The alignment submodule is used to generate question recommendation features based on the text transfer features and the user's historical question text, and to process the question recommendation features to obtain the third text;

[0019] The large model reasoning submodule is used to generate a first prompt based on the first text and the third text, and input the first prompt into the large language model to obtain the recommended question text.

[0020] In the above technical solution, the question recommendation module processes the first text to obtain a first keyword set through a keyword extraction submodule, performs deep semantic inference on the first text to obtain a second text through a deep semantic inference submodule, obtains text transfer features based on the first keyword set, the second text, and user profile information through a diffusion cross-domain submodule, obtains a third text based on the text transfer features and the user's historical question texts through an alignment submodule, and generates a first prompt based on the first and third texts through a large model inference submodule to obtain the recommended question text. Through keyword extraction, deep semantic inference, diffusion cross-domain, text alignment, and large model inference, the recommended question text is obtained. This achieves the goal of maintaining personalized characteristics while also possessing diversity, novelty, uncertainty, and surprise, bringing a better interactive experience to users.

[0021] According to one embodiment of this application, the deep semantic inference submodule is used for:

[0022] The first text is input into the rethink function module of the large language model for deep semantic derivation to obtain the second text.

[0023] In the above technical solution, the first text is input into the rethink function module of the large language model for deep semantic inference to obtain the second text. The second text includes the deep semantic features corresponding to the first text, thus expanding the content of the first text. Furthermore, through deep semantic inference via the rethink function module, the large language model can better understand the context of the first text and adjust the generation of the second text based on the contextual information, making the second text richer in content, closer to real-world situations, and more practical.

[0024] According to one embodiment of this application, the diffusion cross-domain submodule is used for:

[0025] Input the first keyword set, the second text, and the user profile information corresponding to the first input into the diffusion probability model to obtain the text transfer features output by the diffusion probability model.

[0026] In the above technical solution, the first keyword set, the second text, and the user profile information corresponding to the first input are input into the diffusion probability model to obtain the text transfer features output by the diffusion probability model. This allows the establishment of a potential representation distribution and a multi-interest representation of the user, resulting in different interest domains of the user. Based on the user's multiple interest domains, richer and more diverse content can be provided to the user, improving the user's experience and satisfaction.

[0027] According to one embodiment of this application, the alignment submodule is used for:

[0028] Align the text transfer features and the user's historical question text input features with the model to obtain the question recommendation features output by the feature alignment model;

[0029] The recommended features for the problem are discretized and converted into tokens to obtain the third text.

[0030] In the above technical solution, the text transfer features and the user's historical question text are input into the feature alignment model to obtain the question recommendation features output by the feature alignment model. The question recommendation features are discretized and converted into tokens to obtain the third text. This can ensure that the text transfer features are consistent with the real user text features, so that subsequent steps can output high-quality and fluent text, thereby improving the user's experience and satisfaction.

[0031] According to one embodiment of this application, the keyword extraction submodule is used for:

[0032] The first text is input into the keyword extraction model to extract keywords, and the first keyword set output by the keyword extraction model is obtained.

[0033] In the above technical solution, the first text is input into the keyword extraction model for keyword extraction, and the first keyword set is output. The keyword extraction model has powerful representation learning ability and context understanding ability. By encoding and predicting the first text, it realizes automatic keyword extraction, which can better understand the semantic information of the first text and extract a more accurate and comprehensive first keyword set.

[0034] According to one embodiment of this application, the hardware detection module includes:

[0035] Sensor modules are used to collect real-time data from refrigeration equipment;

[0036] The anomaly detection submodule is used to perform anomaly detection on the real-time data based on historical data stored in the background or user-preset data. If anomaly data is detected, anomaly information is generated and sent to the control module.

[0037] In the above technical solution, the hardware detection module collects real-time data from the refrigeration equipment through the sensor module and sends it to the anomaly detection submodule. The anomaly detection submodule performs anomaly detection on the real-time data based on historical data stored in the background or user-preset data. When anomaly data is detected, anomaly information is generated and sent to the control module, realizing real-time monitoring of environmental data of the refrigeration equipment and early warning of anomaly information. This enables users to deal with abnormal situations in a timely manner and improves the user experience of using the refrigeration equipment.

[0038] According to one embodiment of this application, the control module is further configured to:

[0039] Upon receiving the abnormal information, the problem recommendation module is triggered to run.

[0040] In the above technical solution, after the control module receives the abnormal information, it triggers the problem recommendation module to run, which informs the user of the abnormal information and provides related questions and answers, thereby improving the user's experience and satisfaction.

[0041] According to one embodiment of this application, the control module is further configured to:

[0042] If the abnormal information pertains to a topic that has not appeared in the user's history, the interaction module is controlled to output the abnormal information.

[0043] And / or,

[0044] The interactive module is controlled to output the recommended question text within a preset time period, which is a time period specified by the user.

[0045] In the above technical solution, the control module controls the interaction module to output abnormal information when the abnormal information has not appeared in the user's history, or outputs problem recommendation text within a preset time period. By outputting abnormal information and / or problem recommendation text according to certain output rules, the needs of different users can be met, and the user experience and satisfaction can be improved.

[0046] According to one embodiment of this application, the interaction module includes: a data acquisition submodule, a speech-to-text submodule, a text-to-speech submodule, and a speech playback submodule.

[0047] The data acquisition submodule is used to receive the first voice data;

[0048] The speech-to-text submodule is used to convert the first speech data to obtain the first text, and send the first text to the control module;

[0049] The text-to-speech submodule is used to convert the feedback text to obtain second speech data;

[0050] The voice playback submodule is used to play the second voice data;

[0051] The text-to-speech submodule is also used to convert the abnormal information to obtain third speech data, and / or to convert the recommended question text to obtain fourth speech data;

[0052] The voice playback submodule is also used to play the third voice data and / or the fourth voice data.

[0053] In the above technical solution, the interaction module receives the first voice data input by the user through the data acquisition submodule, converts the first voice data into first text through the speech-to-text submodule and sends it to the control module, converts the feedback text into second voice data through the text-to-speech submodule, converts abnormal information into third voice data, converts recommended question text into fourth voice data, and plays the second, third, and / or fourth voice data through the voice playback submodule. The interaction module combines technologies such as voice processing, speech-to-text, and text-to-speech to realize voice interaction between the user and the cooling equipment, which can improve the user experience and provide real-time feedback and personalized services.

[0054] Secondly, this application provides an active question-answering method, which includes:

[0055] The interaction module receives the first input, processes the first input, and obtains the first text;

[0056] The control module sends the first text to the question-and-answer response module to determine the feedback text, and triggers the hardware detection module and the question recommendation module to run based on the first text;

[0057] The question-and-answer response module determines the feedback text based on the first text;

[0058] The hardware detection module collects hardware sensing data, performs abnormal data detection based on the hardware sensing data, generates abnormal information when abnormal data is detected, and sends the abnormal information to the control module.

[0059] The question recommendation module generates recommended question text based on the first text and sends the recommended question text to the control module;

[0060] The control module controls the interaction module to output the feedback text, as well as the error information and / or recommended question text;

[0061] The interaction module outputs the feedback text, as well as the error information and / or recommended question text.

[0062] In the above technical solution, the interaction module receives the first input, processes it to obtain the first text, and the control module sends the first text to the question-and-answer response module to determine the feedback text. Based on the first text, the control module triggers the operation of the hardware detection module and the question recommendation module. The hardware detection module collects hardware sensing data to detect abnormal data. If abnormal data is detected, it generates abnormal information and sends it to the control module. The question recommendation module generates recommended text based on the first text and sends it to the control module. The control module controls the interaction module to output the feedback text, abnormal information, and / or recommended question text. By integrating the question recommendation module and the hardware detection module, while providing normal feedback to the feedback text corresponding to the first input, it also realizes abnormal warning and proactive question recommendation. The question-and-answer mode is diverse, the application scenarios are wide, and the user experience is improved.

[0063] Thirdly, this application provides a refrigeration device that includes the active question-and-answer system as described in the first aspect above.

[0064] In the above technical solution, the active question-and-answer system included in the refrigeration equipment receives and processes the first input through the interaction module to obtain the first text. The control module sends the first text to the question-and-answer response module to determine the feedback text, and triggers the hardware detection module and the question recommendation module to run based on the first text. The hardware detection module collects hardware sensing data to detect abnormal data. If abnormal data is detected, it generates abnormal information and sends it to the control module. The question recommendation module generates recommended text based on the first text and sends it to the control module. The control module controls the interaction module to output the feedback text, abnormal information, and / or recommended question text. By integrating the question recommendation module and the hardware detection module, abnormal warnings and active question recommendations are achieved while providing normal feedback to the feedback text corresponding to the first input. The question-and-answer mode is diverse, the application scenarios are wide, and the user experience is improved.

[0065] Fourthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the active question-answering method as described in the second aspect above.

[0066] Fifthly, this application provides a chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the active question-and-answer method as described in the second aspect.

[0067] In a sixth aspect, this application provides a computer program product, including a computer program that, when executed by a processor, implements the active question-answering method as described in the second aspect above.

[0068] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0069] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0070] Figure 1 This is one of the structural schematic diagrams of the active question-answering system provided in the embodiments of this application;

[0071] Figure 2 This is a schematic diagram of the structure of the interaction module provided in the embodiments of this application;

[0072] Figure 3 This is a schematic diagram illustrating the implementation principle of the control module provided in the embodiments of this application;

[0073] Figure 4 This is a schematic diagram illustrating the implementation principle of the hardware detection module provided in the embodiments of this application;

[0074] Figure 5 This is a schematic diagram illustrating the implementation principle of the problem recommendation module provided in this application embodiment;

[0075] Figure 6 This is a second schematic diagram of the structure of the proactive question-answering system provided in the embodiments of this application;

[0076] Figure 7 This is a flowchart of an active question-answering method provided in an embodiment of this application;

[0077] Figure 8 This is a schematic diagram of the structure of a refrigeration device including an active question-and-answer system provided in an embodiment of this application;

[0078] Figure 9 This is a schematic diagram of an electronic device provided in an embodiment of this application.

[0079] Explanation of reference numerals in the attached figures:

[0080] 1: Active question-and-answer system; 10: Interaction module; 20: Control module;

[0081] 30: Question and Answer Module; 40: Hardware Detection Module; 50: Question Recommendation Module;

[0082] 101: Data Acquisition Submodule; 102: Speech-to-Text Submodule; 103: Text-to-Speech Submodule;

[0083] 104: Voice playback submodule; 80: Refrigeration equipment; 900: Electronic equipment;

[0084] 901: Processor; 902: Memory. Detailed Implementation

[0085] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0086] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0087] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.

[0088] The active question-answering system, method, and refrigeration equipment provided in this application will be described in detail below with reference to the accompanying drawings and through specific embodiments and application scenarios.

[0089] Among them, the active question-answering method of the active question-answering system can be applied to the terminal, and can be executed by the hardware or software in the terminal.

[0090] Figure 1 This is one of the structural schematic diagrams of the active question-answering system provided in the embodiments of this application. For example... Figure 1 As shown, the proactive question-answering system 1 includes: an interaction module 10, a control module 20, a question-answering response module 30, a hardware detection module 40, and a question recommendation module 50.

[0091] The interaction module 10 is used to receive the first input, process the first input, and obtain the first text;

[0092] The control module 20 is used to send the first text to the question-and-answer response module 30 to determine the feedback text, and to trigger the hardware detection module 40 and the question recommendation module 50 to run based on the first text;

[0093] The question-and-answer response module 30 is used to determine the feedback text based on the first text;

[0094] The hardware detection module 40 is used to collect hardware sensing data, perform abnormal data detection based on the hardware sensing data, generate abnormal information when abnormal data is detected, and send the abnormal information to the control module 20.

[0095] The question recommendation module 50 is used to generate recommended question text based on the first text and send the recommended question text to the control module 20;

[0096] The control module 20 is also used to control the interaction module 10 to output feedback text, as well as error information and / or recommended question text;

[0097] The interaction module 10 is also used to output feedback text, as well as error information and / or recommended question text.

[0098] Figure 2 This is a schematic diagram of the structure of the interactive module provided in the embodiments of this application, such as... Figure 2 As shown, the interaction module 10 may include a data acquisition submodule 101, a speech-to-text submodule 102, a text-to-speech submodule 103, and a speech playback submodule 104.

[0099] Optionally, 101 in the interaction module 10 can use a voice receiver (such as a microphone, mobile phone, etc.) to complete real-time voice information collection. In this case, the first input is the voice command issued by the user, such as "What are some good places to eat in Beijing?", and the voice receiver receives the user's voice command as the first input.

[0100] Optionally, 102 in the interaction module 10 can perform data cleaning on the first input, such as noise reduction, echo removal, and reverberation removal, and then use Whisper to convert the first input into the first text. Whisper is an open-source automatic speech recognition system from OpenAI, which converts the first input into the first text "What are some good places to eat in Beijing".

[0101] Optionally, the first input can also be text-to-speech input via a terminal device or screen device, allowing users to input questions when it is inconvenient for them to speak.

[0102] Optionally, the first input can also be other non-voice inputs, such as gesture input, text input, etc. For example, if the first input is a gesture input, processing the first input includes: performing gesture recognition on the first input and converting the gesture recognition result into first text. For example, if the first input is a text input, processing the first input includes: extracting valid text information from the first input to obtain the first text.

[0103] Figure 3 This is a schematic diagram illustrating the implementation principle of the control module provided in the embodiments of this application, as shown below. Figure 3 As shown, the control module 20 uses Agent Tools to achieve interaction with other modules. Therefore, the control module 20 can also be called the Agent module. Agent Tools can realize functions such as task management, communication tools, resource management, data analysis, information search, and document processing. That is, the control module 20 can control the interaction between the interaction module 10, the question and answer response module 30, the hardware detection module 40, and the question recommendation module 50 in the active question and answer system.

[0104] The control module 20 sends the first text to the question-and-answer response module 30 to determine the feedback text. It can be understood that the question-and-answer response module 30 is a module that can receive the first text and give a corresponding reply based on the first text.

[0105] For example, the control module 20 inputs the first text "What are some good foods in Beijing?" into Agent Tools. Agent Tools then sends the first text to the question-and-answer response module 30. The question-and-answer response module 30 generates feedback text based on the first text, such as "Beijing has Peking duck, candied hawthorn, donkey meat sandwiches, and other delicacies," and returns the feedback text to the control module 20. The control module 20 then controls the interaction module 10 to output the feedback text. Modules 103 and 104 in the interaction module 10 can convert the feedback text into speech and play it to the user.

[0106] It should be noted that the control module 20 will also trigger the hardware detection module 40 and the problem recommendation module 50 to run simultaneously based on the first text.

[0107] Figure 4 This is a schematic diagram illustrating the implementation principle of the hardware detection module provided in this application embodiment, as shown below. Figure 4 As shown, the hardware detection module 40 detects abnormal data by collecting hardware sensing data. When abnormal data is detected, it generates abnormal information and sends the abnormal information to the control module 20.

[0108] Hardware sensing data can be real-time data collected by sensor modules in refrigeration equipment, such as environmental data like temperature, humidity, and odor in the refrigerator or freezer compartment; pressure data from components like the compressor, condenser, and evaporator; or electrical parameters of the refrigeration equipment, such as fluctuations in current and voltage, and refrigerant flow rate.

[0109] For example, the hardware detection module 40 can use a temperature sensor to detect the temperature inside the refrigeration equipment. The temperature sensor monitors the temperature inside the refrigeration equipment in real time and uploads it to the hardware detection module 40. The hardware detection module 40 determines whether there is an anomaly based on the historical data records of the refrigeration equipment. If there are multiple historical data, the average value can be taken. When an anomaly is detected, the hardware detection module 40 generates an anomaly message and sends it to the control module 20. The control module 20 controls the interaction module 10 to convert the anomaly message into voice and play it to the user.

[0110] Optionally, the user can preset a temperature value as a condition for judging whether there is an anomaly. When the temperature sensor detects that the temperature inside the current refrigeration equipment exceeds the preset temperature value, an anomaly message is generated. For example, if the temperature inside the refrigeration equipment is set by the user to 4 to 8 degrees, when the temperature sensor detects that the temperature inside the refrigeration equipment rises to 9 degrees, the hardware detection module 40 generates an anomaly message and sends it to the control module 20. The control module 20 controls 103 and 104 in the interaction module 10 to convert the anomaly message into voice and play it to the user.

[0111] It should be noted that the refrigeration equipment in the embodiments of this application includes refrigerators, freezers and other equipment that can refrigerate. The refrigeration equipment in the embodiments of this application can also be replaced by a cargo container, all of which are within the protection scope of this application.

[0112] To address the issue of poor user experience caused by relying on users' historical voice interaction information, historical life trajectories, or hardware monitoring to implement question-and-answer functions in related technologies, this application proposes setting up a question recommendation module to implement proactive questioning, thereby improving user experience. The question recommendation module generates recommended question text based on a first text. The recommended question text is text generated based on the first text and is used to achieve proactive question-and-answer interaction with the user. In related technologies, the generation of proactive questions mainly revolves around the topics involved in the current user's question, generating questions with similar themes. For example, if a user asks about today's weather, the generated questions will also be weather-related: recommending clothing based on the current weather. Analogous to recommendation systems, which recommend similar items based on the user's browsing history, users will constantly receive recommendations of similar or related items in a short period, but over time, this can lead to user fatigue and boredom. Similarly, question-and-answer systems, based on the user's current or recent questions, continuously generate questions within the same domain (topic or entity information, etc.) as the user. Such proactively generated questions can also trap users in a certain domain without offering new ideas, resulting in a poor user experience. In this embodiment, the recommended question text crosses domains relative to the first text; that is, the recommended question text is a cross-domain text generated based on the first text. Specifically, this embodiment utilizes large language model and diffusion probability model techniques to generate recommended question text based on the first text, helping the question-answering system to proactively propose cross-domain questions.

[0113] Optionally, generating recommendation question text based on the first text includes:

[0114] The first text is subjected to diffusion cross-domain processing to obtain text transfer features, and the text transfer features are subjected to text alignment processing to obtain the third text;

[0115] After processing the first text and the third text according to the preset prompt format, the first prompt is obtained. The first prompt is then input into the large language model to obtain the recommended question text.

[0116] By performing diffusion-crossing processing on the first text, a latent representation distribution and multi-interest representation of the user can be established, resulting in different interest domains of the user. This makes the recommended questions novel, uncertain, and surprising. By performing text alignment processing on the text transfer features, it can be ensured that the text transfer features are consistent with the real user text features. After processing the first and third texts according to the preset prompt specifications and inputting them into the large language model, the resulting recommended question text can maintain the characteristics of user personalization while also having diversity, novelty, uncertainty, and surprise, which can bring a better interactive experience to the user.

[0117] Figure 5This is a schematic diagram illustrating the implementation principle of the problem recommendation module provided in this application embodiment, such as... Figure 5 As shown, the steps of the question recommendation module 50 in generating recommended question text based on the first text can be divided into the following steps:

[0118] 1) The question recommendation module 50 can extract keywords from the first text using the transformer model. For example, if the first text is "What are some good foods in Beijing", the keywords extracted by the transformer model can be: Beijing, good food.

[0119] 2) The question recommendation module 50 can use the rethink function of the Large Language Model (LLM) to obtain the deep semantics of the first text. For example, based on the first text "What are some good places to eat in Beijing?", LLM rethink can deduce that: the current user's location is Beijing, Beijing has delicious food, food is related to location, and food is related to history. Using external knowledge from LLM rethink to supplement the extraction process of deep semantic features of the first text can enhance the ability to extract deep semantic features of the first text.

[0120] 3) The question recommendation module 50 can input the keywords of the first text, user profile information (such as information such as a 23-year-old female) and the text generated by LLM rethink into the diffusion probability model to obtain text transfer features. The diffusion probability model can be used to establish the latent representation distribution of the first text and the multi-interest representation of users.

[0121] 4) The question recommendation module 50 can input user text features into the Alignment Model (AM) and fuse user historical information (e.g., what's the weather like today? What are some tourist attractions in Beijing?). The features output by the Alignment Model are discretized and converted into tokens to generate new text (e.g., historical stories about Beijing). The first text and the generated text are input into the LLM under a system-defined prompt to generate text, such as "What is the history of Beijing?". This text is then sent as the recommendation question text to the control module 20. The control module 20 controls modules 103 and 104 in the interaction module 10 to convert the recommendation question text into speech and play it to the user.

[0122] It is worth noting that the control module 20 can control whether the above-mentioned feedback text, abnormal information, and recommended question text are sent to the user through the interaction module 10. The user can set to receive a reply within a certain time period, such as from 8 am to 8 pm. During this time period, the control module 20 controls the above-mentioned feedback text, abnormal information, and recommended question text to be converted into speech and played to the user through the interaction module 10.

[0123] Figure 6 This is a second schematic diagram of the structure of the active question-answering system provided in the embodiments of this application, as shown below. Figure 6 As shown, interaction module 101 receives the user's voice as the first input. Interaction module 1011 cleans the first input and uses Whisper or other speech-to-text tools to convert the first input into first text, which is then sent to control module 20 (i.e., Agent module). Agent module controls question-and-answer response module 30 to determine the feedback text and converts the feedback text into speech and plays it to the user through interaction module 1010. At the same time, hardware detection module 40 and question recommendation module 50 are triggered. Hardware detection module 40 converts the detected abnormal information into speech and plays it to the user through interaction modules 103 and 104. Question recommendation module 50 generates recommended question text based on the first text and converts the recommended question text into speech and plays it to the user through interaction module 10100.

[0124] The proactive question-answering system provided in this application embodiment receives a first input through an interaction module, processes it to obtain a first text, and sends the first text to a question-answering response module to determine the feedback text. Based on the first text, the control module triggers the operation of a hardware detection module and a question recommendation module. The hardware detection module collects hardware sensing data to detect abnormal data. If abnormal data is detected, it generates abnormal information and sends it to the control module. The question recommendation module generates recommended text based on the first text and sends it to the control module. The control module controls the interaction module to output the feedback text, abnormal information, and / or recommended question text. By integrating the question recommendation module and the hardware detection module, the system achieves both normal feedback to the feedback text corresponding to the first input and abnormal warning and proactive question recommendation. It offers diverse question-answering modes, wide application scenarios, and enhances the user experience.

[0125] In one embodiment of this application, the question recommendation module includes:

[0126] The keyword extraction submodule is used to extract keywords from the first text to obtain a first keyword set.

[0127] The deep semantic derivation submodule is used to perform deep semantic derivation based on the first text to obtain the second text, wherein the second text includes deep semantic features;

[0128] The diffusion cross-domain submodule is used to generate text transfer features based on the first keyword set, the second text, and the user profile information corresponding to the first input;

[0129] The alignment submodule is used to generate question recommendation features based on the text transfer features and the user's historical question text, and to process the question recommendation features to obtain the third text;

[0130] The large model reasoning submodule is used to generate a first prompt based on the first text and the third text, and input the first prompt into the large language model to obtain the recommended question text.

[0131] In one embodiment of this application, the keyword extraction submodule is used for:

[0132] The first text is input into the keyword extraction model to extract keywords, and the first keyword set output by the keyword extraction model is obtained.

[0133] It should be noted that keywords can be words that appear frequently in the first text, or words or phrases that are of great significance in the first text.

[0134] Optionally, the keyword extraction model can be a Transformer model, a Term Frequency-Inverse Document Frequency (TF-IDF) model, a TextRank model, a Bidirectional Encoder Representations from Transformers (BERT) model, a RoBERTa model, a GPT model, etc., and this application embodiment does not limit it.

[0135] In the above technical solution, the first text is input into the keyword extraction model for keyword extraction, and the first keyword set is output. The keyword extraction model has powerful representation learning ability and context understanding ability. By encoding and predicting the first text, it realizes automated keyword extraction, which can better understand the semantic information of the first text and extract a more accurate and comprehensive first keyword set.

[0136] In one embodiment of this application, the deep semantic inference submodule is used for:

[0137] The first text is input into the rethink function module of the large language model for deep semantic derivation to obtain the second text.

[0138] It should be noted that the rethink function module generally refers to the ability of a large language model to reconsider the first input text and perform in-depth semantic deduction to obtain the second text.

[0139] Optionally, the first text can be input into the rethink function module of the Large Language Model (LLM) for deep semantic derivation to obtain the second text.

[0140] It is worth noting that large language models can also employ semantic compression large language models LLM with decoder-ONLY, encoder-ONLY, and encoder-decoder, such as graph neural networks, attention mechanisms, transformer models and their variants or improvements, distillation network LLM, latent / diffusion model LLM, U-net network and other deep network models, as well as Gaussian mixture deep neural network models.

[0141] In the above technical solution, the first text is input into the rethink function module of the large language model for deep semantic inference to obtain the second text. The second text includes the deep semantic features corresponding to the first text, thus expanding the content of the first text. Furthermore, through deep semantic inference via the rethink function module, the large language model can better understand the context of the first text and adjust the generation of the second text based on the contextual information, making the second text richer in content, closer to real-world situations, and more practical.

[0142] In one embodiment of this application, the diffusion cross-domain submodule is used for:

[0143] Input the first keyword set, the second text, and the user profile information corresponding to the first input into the diffusion probability model to obtain the text transfer features output by the diffusion probability model.

[0144] It should be noted that the user profile information corresponding to the first input refers to the description and summary of user characteristics, interests, behaviors, etc., formed by collecting, organizing, and analyzing various user data. Inputting user profile information into the diffusion probability model can better understand user needs and improve user experience and satisfaction.

[0145] Optionally, user preferences, likes, and feedback can be collected through questionnaires to obtain user profile information.

[0146] Optionally, data mining techniques such as cluster analysis and association rule mining can be used to analyze and mine large amounts of user data to discover the user behavior patterns and characteristics hidden behind the data, thereby obtaining user profile information.

[0147] Optionally, machine learning models, such as classification and regression algorithms, can be used to learn user characteristics and behavioral patterns from historical data, thereby obtaining user profile information.

[0148] It is worth noting that the Diffusion Probabilistic Model (DPM) outputs text transfer features based on the first keyword set, the second text, and user profile information, which can be divided into the following steps:

[0149] 1) Data preprocessing: First, DPM preprocesses the first keyword set and the second text, including word segmentation, stop word removal, stemming, etc., for subsequent processing.

[0150] 2) User profile information integration: DPM integrates user profile information into the input data, for example, by encoding user profile information into vectors and connecting or concatenating them with the input data.

[0151] 3) Construct a diffusion probability model: Design and train a DPM. The input of the DPM includes a preprocessed first keyword set, a second text, and a vector integrating user profile information. The output of the DPM is the text transfer feature, which can be a probability distribution or a continuous value vector.

[0152] 4) Establishment of latent representation distribution: Using the obtained text transfer features, clustering algorithms or dimensionality reduction techniques, such as principal component analysis and t-distributed stochastic neighbor embedding (t-SNE), can be used to construct the latent representation distribution of the output text. This can map the output text into a low-dimensional space while preserving the semantic similarity of the output text.

[0153] 5) Multi-interest representation of users: Based on users' historical behavior data and text transfer features output by DPM, clustering or classification models can be used to build multi-interest representations of users.

[0154] In the above technical solution, the first keyword set, the second text, and the user profile information corresponding to the first input are input into the diffusion probability model to obtain the text transfer features output by the diffusion probability model. This allows the establishment of a potential representation distribution and a multi-interest representation of the user, resulting in different interest domains of the user. Based on the user's multiple interest domains, richer and more diverse content can be provided to the user, improving the user's experience and satisfaction.

[0155] In one embodiment of this application, the alignment submodule is used for:

[0156] Align the text transfer features and the user's historical question text input features with the model to obtain the question recommendation features output by the feature alignment model;

[0157] The recommended features for the problem are discretized and converted into tokens to obtain the third text.

[0158] It is easy to understand that by inputting the text transfer features and the user's historical question text into the feature alignment model (AM), the question recommendation features output by the alignment model can ensure that the text transfer features and the user's historical question text are consistent.

[0159] Optionally, the alignment model can be a Siamese network model, which can learn the similarity or distance between text transfer features and the user's historical question text.

[0160] Alternatively, the alignment model can be a Siamese self-attention network (Siamese Network) model, which is a Siamese network model that incorporates a self-attention mechanism. It can better capture global dependencies in a text sequence when aligning text features.

[0161] Optionally, the alignment model can be a pre-trained language model, such as BERT or GPT, which can be used to align text features. By fine-tuning these models, they can be made to align and match text transfer features with the user's historical question text features.

[0162] Optionally, the alignment model can be a Transformer Model, a powerful neural network architecture widely used in natural language processing tasks. They can be used to align text features, capture the complex relationships between text transition features and the user's historical question text, and generate corresponding question recommendation features.

[0163] Optionally, the alignment model can be a cross-modal alignment model. If the input text features come from different modalities (such as text and images), a cross-modal alignment model can be used to learn the correspondence between different modalities.

[0164] Optionally, the alignment model can be a transfer learning model, which can share knowledge across different domains, thereby helping to align text features from different sources.

[0165] The alignment submodule discretizes the problem recommendation features, converting continuous problem recommendation features into a set of discrete values.

[0166] Optionally, discretization can be achieved by dividing a continuous range of values ​​into discrete intervals or by using a clustering algorithm. For example, the range of values ​​for the recommendation features can be divided into several intervals of equal width or frequency, and then each interval can be mapped to a discrete value.

[0167] The alignment submodule converts the discretized question recommendation features into tokens to obtain third-party text. The question recommendation features can be represented as a sequence of predefined tokens or labels. These tokens typically represent specific semantics or meanings. For example, words from a vocabulary can be used as tokens, or a pre-trained word embedding model can be used to map words into a continuous vector space. These vectors are then used as tokens to obtain the third-party text.

[0168] In the above technical solution, the text transfer features and the user's historical question text are input into the feature alignment model to obtain the question recommendation features output by the feature alignment model. The question recommendation features are discretized and converted into tokens to obtain the third text. This can ensure that the text transfer features are consistent with the real user text features, so that subsequent steps can output high-quality and fluent text, thereby improving the user's experience and satisfaction.

[0169] After obtaining the third text, the large model reasoning submodule can process the first and third texts according to the preset prompt generation specifications to generate the first prompt. The first prompt is then input into the large language model to obtain the recommended question text.

[0170] In the above technical solution, the question recommendation module processes the first text to obtain a first keyword set through a keyword extraction submodule, performs deep semantic inference on the first text to obtain a second text through a deep semantic inference submodule, obtains text transfer features based on the first keyword set, the second text, and user profile information through a diffusion cross-domain submodule, obtains a third text based on the text transfer features and the user's historical question texts through an alignment submodule, and generates a first prompt based on the first and third texts through a large model inference submodule to obtain the recommended question text. Through keyword extraction, deep semantic inference, diffusion cross-domain, text alignment, and large model inference, the recommended question text is obtained. This achieves the goal of maintaining personalized characteristics while also possessing diversity, novelty, uncertainty, and surprise, bringing a better interactive experience to users.

[0171] In one embodiment of this application, the hardware detection module includes:

[0172] Sensor modules are used to collect real-time data from refrigeration equipment;

[0173] The anomaly detection submodule is used to perform anomaly detection on the real-time data based on historical data stored in the background or user-preset data. If anomaly data is detected, anomaly information is generated and sent to the control module.

[0174] Real-time data can be environmental data such as temperature, humidity, and odor in the refrigerator or freezer compartment, pressure data of components such as compressors, condensers, and evaporators, or electrical parameters of refrigeration equipment, such as fluctuations in current and voltage, and refrigerant flow rate.

[0175] Optionally, the sensor module includes at least one of the following: temperature sensor, humidity sensor, odor sensor, pressure sensor, current sensor, voltage sensor, and flow sensor.

[0176] Temperature sensors, including thermistors, thermocouples, and resistance temperature probes, are used to monitor temperature changes within refrigeration equipment. The temperature sensors send the collected data to an anomaly detection submodule. This submodule performs anomaly detection based on historical data stored in the background or user-preset data. If anomalies are detected, it generates an anomaly message and sends it to the control module.

[0177] Humidity sensors, including capacitive and resistive humidity sensors, are used to monitor humidity changes within refrigeration equipment. The humidity sensor sends the collected data to an anomaly detection submodule. This submodule performs anomaly detection based on historical data stored in the background or user-preset data. If anomalies are detected, it generates an anomaly message and sends it to the control module.

[0178] Odor sensors, including semiconductor gas sensors and artificial olfaction sensors, are used to monitor food spoilage within refrigeration equipment. The odor sensors send the collected data to an anomaly detection submodule. This submodule performs anomaly detection based on historical data stored in the background or user-preset data. If anomalies are detected, it generates an anomaly message and sends it to the control module.

[0179] Pressure sensors include piezoelectric, resistive, capacitive, and resonant pressure sensors, which are used to monitor pressure changes in refrigeration equipment. The pressure sensors send the collected data to the anomaly detection submodule. The anomaly detection submodule performs anomaly detection on the real-time data based on historical data stored in the background or user-preset data. When anomaly data is detected, anomaly information is generated and sent to the control module.

[0180] Current sensors, including resistive current sensors, current transformers, and inductive current sensors, are used to monitor current changes in refrigeration equipment. The current sensors send the collected data to the anomaly detection submodule. The anomaly detection submodule performs anomaly detection on real-time data based on historical data stored in the background or user-preset data. If anomaly data is detected, anomaly information is generated and sent to the control module.

[0181] Voltage sensors, including voltage transformers, capacitive voltage sensors, and inductive voltage sensors, are used to monitor voltage changes in refrigeration equipment. The voltage sensors send the collected data to the anomaly detection submodule. The anomaly detection submodule performs anomaly detection on the real-time data based on historical data stored in the background or user-preset data. If anomaly data is detected, anomaly information is generated and sent to the control module.

[0182] Flow sensors, including electromagnetic flow sensors and ultrasonic flow sensors, are used to monitor changes in flow rate in refrigeration equipment. The flow sensors send the collected data to the anomaly detection submodule. The anomaly detection submodule performs anomaly detection on the real-time data based on historical data stored in the background or user-preset data. If anomaly data is detected, anomaly information is generated and sent to the control module.

[0183] It should be noted that the above sensor modules can be used individually or in combination to achieve comprehensive monitoring and control of various parameters in the refrigeration equipment.

[0184] In the above technical solution, the hardware detection module collects real-time data from the refrigeration equipment through the sensor module and sends it to the anomaly detection submodule. The anomaly detection submodule performs anomaly detection on the real-time data based on historical data stored in the background or user-preset data. When anomaly data is detected, anomaly information is generated and sent to the control module, realizing real-time monitoring of environmental data of the refrigeration equipment and early warning of anomaly information. This enables users to deal with abnormal situations in a timely manner and improves the user experience of using the refrigeration equipment.

[0185] In one embodiment of this application, the control module is further configured to:

[0186] Upon receiving the abnormal information, the problem recommendation module is triggered to run.

[0187] It is worth noting that after receiving the abnormal information sent by the hardware detection module, the control module triggers the problem recommendation module to run. Based on the abnormal information text, the problem recommendation module uses the keyword extraction submodule to extract keywords from the abnormal information text to obtain the first keyword set, uses the deep semantic inference submodule to perform deep semantic inference on the abnormal information text to obtain the second text, uses the diffusion cross-domain submodule to obtain text transfer features based on the first keyword set, the second text, and user profile information, uses the alignment submodule to obtain the third text based on the text transfer features and the user's historical problem text, and uses the large model inference submodule to generate the first prompt based on the abnormal information text and the third text to obtain the abnormal information recommendation text.

[0188] In the above technical solution, after the control module receives the abnormal information, it triggers the problem recommendation module to run, which informs the user of the abnormal information and provides related questions and answers, thereby improving the user's experience and satisfaction.

[0189] In one embodiment of this application, the control module is further configured to:

[0190] If the abnormal information pertains to a topic that has not appeared in the user's history, the interaction module is controlled to output the abnormal information.

[0191] And / or,

[0192] The interactive module is controlled to output the recommended question text within a preset time period, which is a time period specified by the user.

[0193] Optionally, when the abnormal information is a topic that has not appeared in the user's history, the control interaction module outputs the abnormal information and informs the user that this is the first time the abnormal information has occurred.

[0194] For example, an anomaly message might indicate that the pressure in the refrigeration equipment has suddenly risen above the normal range, a situation that has never occurred before in the user's history. In this case, the control module will instruct the interaction module to output an anomaly message, such as through the user interface or an alarm, to notify the user that this is the first time this anomaly has occurred. The user may then take this seriously, promptly inspect the equipment, or contact maintenance personnel to understand the root cause of the problem and take steps to resolve the issue.

[0195] Optionally, the control module controls the interaction module to output recommended question text within a preset time period. The preset time period is a time period specified by the user. For example, if the user does not want to be disturbed at night, the preset time period is set to 10:00 AM to 7:00 PM. The control module controls the interaction module to only output recommended question text within the time range of 10:00 AM to 7:00 PM, and will not output recommended question text at other times.

[0196] In the above technical solution, the control module controls the interaction module to output abnormal information when the abnormal information has not appeared in the user's history, or outputs problem recommendation text within a preset time period. By outputting abnormal information and / or problem recommendation text according to certain output rules, the needs of different users can be met, and the user experience and satisfaction can be improved.

[0197] In one embodiment of this application, the interaction module includes: a data acquisition submodule, a speech-to-text submodule, a text-to-speech submodule, and a speech playback submodule.

[0198] The data acquisition submodule is used to receive the first voice data;

[0199] The speech-to-text submodule is used to convert the first speech data to obtain the first text, and send the first text to the control module;

[0200] The text-to-speech submodule is used to convert the feedback text to obtain second speech data;

[0201] The voice playback submodule is used to play the second voice data;

[0202] The text-to-speech submodule is also used to convert the abnormal information to obtain third speech data, and / or to convert the recommended question text to obtain fourth speech data;

[0203] The voice playback submodule is also used to play the third voice data and / or the fourth voice data.

[0204] In this embodiment of the application, the data acquisition submodule can receive first voice data input by the user. The data acquisition submodule can be a microphone, sound processor, speaker, etc. The first voice data is, for example, the user saying "What should I eat today?" or "How is the weather in Beijing?"

[0205] Optionally, the speech-to-text submodule converts the first voice data input by the user into the first text. The speech-to-text submodule can be an embedded speech recognition module, which can be integrated into the refrigeration equipment to realize the speech-to-text function.

[0206] Optionally, the speech-to-text submodule can be a cloud-based speech recognition service module that implements speech-to-text functionality through an API interface.

[0207] It is easy to understand that the speech-to-text submodule sends the first text to the control module, the control module processes the first text to obtain feedback text, and sends it to the text-to-speech submodule of the interaction module. The text-to-speech submodule can be an embedded speech synthesis module. The text-to-speech submodule can convert the feedback text into second speech data, and can also convert the above-mentioned abnormal information into third speech data and / or convert the above-mentioned recommendation question text into fourth speech data.

[0208] In this embodiment, the voice playback submodule can play the second voice data, the third voice data, and / or the fourth voice data to the user. The voice playback submodule is, for example, a speaker, a speaker box, or the voice playback module of an embedded system.

[0209] In the above technical solution, the interaction module receives the first voice data input by the user through the data acquisition submodule, converts the first voice data into first text through the speech-to-text submodule and sends it to the control module, converts the feedback text into second voice data through the text-to-speech submodule, converts abnormal information into third voice data, converts recommended question text into fourth voice data, and plays the second, third, and / or fourth voice data through the voice playback submodule. The interaction module combines technologies such as speech recognition, text processing, and speech synthesis to realize voice interaction between the user and the cooling equipment, which can improve the user experience, simplify the operation process, and provide real-time feedback and personalized services.

[0210] This application also provides an active question-and-answer method. Figure 7 This is a flowchart of an active question-answering method provided in an embodiment of this application, such as... Figure 7 As shown, the active question answering method of the active question answering system includes steps 710, 720, 730, 740, 750, 770 and 770.

[0211] Step 710: The interaction module receives the first input, processes the first input, and obtains the first text;

[0212] Step 720: The control module sends the first text to the question-and-answer response module to determine the feedback text, and triggers the hardware detection module and the question recommendation module to run based on the first text;

[0213] Step 730: The question-and-answer response module determines the feedback text based on the first text;

[0214] Step 740: The hardware detection module collects hardware sensing data, performs abnormal data detection based on the hardware sensing data, generates abnormal information when abnormal data is detected, and sends the abnormal information to the control module.

[0215] Step 750: The question recommendation module generates recommended question text based on the first text and sends the recommended question text to the control module;

[0216] Step 760: The control module controls the interaction module to output the feedback text, as well as the exception information and / or recommended question text;

[0217] Step 770: The interaction module outputs the feedback text, as well as the exception information and / or recommended question text.

[0218] In the above technical solution, the interaction module receives the first input, processes it to obtain the first text, and the control module sends the first text to the question-and-answer response module to determine the feedback text. Based on the first text, the control module triggers the operation of the hardware detection module and the question recommendation module. The hardware detection module collects hardware sensing data to detect abnormal data. If abnormal data is detected, it generates abnormal information and sends it to the control module. The question recommendation module generates recommended text based on the first text and sends it to the control module. The control module controls the interaction module to output the feedback text, abnormal information, and / or recommended question text. By integrating the question recommendation module and the hardware detection module, while providing normal feedback to the feedback text corresponding to the first input, it also realizes abnormal warning and proactive question recommendation. The question-and-answer mode is diverse, the application scenarios are wide, and the user experience is improved.

[0219] Optionally, generating recommendation question text based on the first text includes:

[0220] Keyword extraction is performed on the first text to obtain a first keyword set;

[0221] Based on the first text, a second text is obtained by performing deep semantic derivation, and the second text includes deep semantic features.

[0222] Text transfer features are generated based on the first keyword set, the second text, and the user profile information corresponding to the first input.

[0223] Based on the text transfer features and the user's historical question text, question recommendation features are generated, and the question recommendation features are processed to obtain the third text;

[0224] A first prompt is generated based on the first text and the third text. The first prompt is then input into a large language model to obtain the recommended question text.

[0225] Optionally, the step of extracting keywords from the first text to obtain a first keyword set includes:

[0226] The first text is input into the keyword extraction model to extract keywords, and the first keyword set output by the keyword extraction model is obtained.

[0227] Optionally, the step of performing deep semantic deduction based on the first text to obtain the second text includes:

[0228] The first text is input into the rethink function module of the large language model for deep semantic derivation to obtain the second text.

[0229] Optionally, generating text transfer features based on the first keyword set, the second text, and the user profile information corresponding to the first input includes:

[0230] Input the first keyword set, the second text, and the user profile information corresponding to the first input into the diffusion probability model to obtain the text transfer features output by the diffusion probability model.

[0231] Optionally, the step of generating question recommendation features based on the text transfer features and the user's historical question text, and processing the question recommendation features to obtain a third text, includes:

[0232] Align the text transfer features and the user's historical question text input features with the model to obtain the question recommendation features output by the feature alignment model;

[0233] The recommended features for the problem are discretized and converted into tokens to obtain the third text.

[0234] Optionally, the hardware detection module collects hardware sensing data, performs abnormal data detection based on the hardware sensing data, and generates abnormal information when abnormal data is detected, including:

[0235] Collect real-time data from refrigeration equipment;

[0236] Based on historical data stored in the background or user-preset data, anomaly detection is performed on the real-time data, and anomaly information is generated when anomaly data is detected.

[0237] Optionally, the method further includes:

[0238] Upon receiving the abnormal information, the hardware detection module triggers the problem recommendation module to run.

[0239] Optionally, the method further includes:

[0240] If the abnormal information pertains to a topic that has not appeared in the user's history, the control module controls the interaction module to output the abnormal information.

[0241] And / or,

[0242] The control module controls the interaction module to output the recommended question text within a preset time period, which is a time period specified by the user.

[0243] Optionally, receiving the first input and processing the first input to obtain the first text includes:

[0244] Receive the first voice data;

[0245] The first speech data is converted to obtain the first text;

[0246] Optionally, the output of the feedback text, as well as the exception information and / or recommended question text, includes:

[0247] The feedback text is converted to obtain second speech data;

[0248] Play the second voice data;

[0249] The abnormal information is converted to obtain third voice data, and / or the recommended question text is converted to obtain fourth voice data;

[0250] Play the third and / or fourth voice data.

[0251] For an understanding of the above-mentioned proactive question answering method, please refer to the description in the aforementioned proactive question answering system embodiments, which can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0252] This application also provides a refrigeration device that includes the aforementioned active question-and-answer system. Figure 8 This is a schematic diagram of the structure of a refrigeration device including an active question-and-answer system provided in an embodiment of this application. For example... Figure 8 As shown, the refrigeration equipment 80 includes an active question-and-answer system 1.

[0253] In the above technical solution, the active question-and-answer system included in the refrigeration equipment receives and processes the first input through the interaction module to obtain the first text. The control module sends the first text to the question-and-answer response module to determine the feedback text, and triggers the hardware detection module and the question recommendation module to run based on the first text. The hardware detection module collects hardware sensing data to detect abnormal data. If abnormal data is detected, it generates abnormal information and sends it to the control module. The question recommendation module generates recommended text based on the first text and sends it to the control module. The control module controls the interaction module to output the feedback text, abnormal information, and / or recommended question text. By integrating the question recommendation module and the hardware detection module, abnormal warnings and active question recommendations are achieved while providing normal feedback to the feedback text corresponding to the first input. The question-and-answer mode is diverse, the application scenarios are wide, and the user experience is improved.

[0254] For an understanding of the above-mentioned refrigeration equipment, please refer to the description in the aforementioned active question-and-answer system embodiment, which can achieve the same technical effect. To avoid repetition, it will not be repeated here.

[0255] The control module in this application embodiment can be an electronic device or a component within an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), television (TV), ATM, or self-service machine, etc. This application embodiment does not specifically limit the scope of the electronic device.

[0256] The control module in this embodiment can be a device with an operating system. This operating system can be Microsoft Windows, Android, iOS, or other possible operating systems; this embodiment does not specifically limit the specific operating system.

[0257] In some embodiments, such as Figure 9 As shown, this application embodiment also provides an electronic device 900, including a processor 901, a memory 902, and a computer program stored in the memory 902 and executable on the processor 901. When the program is executed by the processor 901, it implements the various processes of the above-described active question-answering method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0258] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0259] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described active question-answering method embodiments and achieves the same technical effect. To avoid repetition, it will not be described again here.

[0260] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0261] This application also provides a computer program product, including a computer program that implements an active question-and-answer method when executed by a processor.

[0262] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0263] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described active question-answering method embodiments and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0264] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0265] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0266] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0267] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0268] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0269] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. An active question-answering system, characterized in that, include: The module includes an interaction module, a control module, a question-and-answer module, a hardware detection module, and a question recommendation module. The interaction module is used to receive a first input, process the first input, and obtain a first text; The control module is used to send the first text to the question-and-answer response module to determine the feedback text, and to trigger the hardware detection module and the question recommendation module to run based on the first text; The question-and-answer response module is used to determine the feedback text based on the first text; The hardware detection module is used to collect hardware sensing data, perform abnormal data detection based on the hardware sensing data, generate abnormal information when abnormal data is detected, and send the abnormal information to the control module. The question recommendation module is used to generate recommended question text based on the first text and send the recommended question text to the control module; The control module is also used to control the interaction module to output the feedback text, as well as the abnormal information and / or recommended question text; The interaction module is also used to output the feedback text, as well as the error information and / or recommended question text.

2. The active question-answering system according to claim 1, characterized in that, The problem recommendation module includes: The keyword extraction submodule is used to extract keywords from the first text to obtain a first keyword set. The deep semantic derivation submodule is used to perform deep semantic derivation based on the first text to obtain the second text, wherein the second text includes deep semantic features; The diffusion cross-domain submodule is used to generate text transfer features based on the first keyword set, the second text, and the user profile information corresponding to the first input; The alignment submodule is used to generate question recommendation features based on the text transfer features and the user's historical question text, and to process the question recommendation features to obtain the third text; The large model reasoning submodule is used to generate a first prompt based on the first text and the third text, and input the first prompt into the large language model to obtain the recommended question text.

3. The active question-answering system according to claim 2, characterized in that, The deep semantic inference submodule is used for: The first text is input into the rethink function module of the large language model for deep semantic derivation to obtain the second text.

4. The active question-answering system according to claim 2, characterized in that, The diffusion cross-domain submodule is used for: Input the first keyword set, the second text, and the user profile information corresponding to the first input into the diffusion probability model to obtain the text transfer features output by the diffusion probability model.

5. The active question-answering system according to claim 2, characterized in that, The alignment submodule is used for: Align the text transfer features and the user's historical question text input features with the model to obtain the question recommendation features output by the feature alignment model; The recommended features for the problem are discretized and converted into tokens to obtain the third text.

6. The active question-answering system according to claim 2, characterized in that, The keyword extraction submodule is used for: The first text is input into the keyword extraction model to extract keywords, and the first keyword set output by the keyword extraction model is obtained.

7. The active question-answering system according to claim 1, characterized in that, The hardware detection module includes: Sensor modules are used to collect real-time data from refrigeration equipment; The anomaly detection submodule is used to perform anomaly detection on the real-time data based on historical data stored in the background or user-preset data. If anomaly data is detected, anomaly information is generated and sent to the control module.

8. The active question-answering system according to claim 7, characterized in that, The control module is also used for: Upon receiving the abnormal information, the problem recommendation module is triggered to run.

9. The active question-answering system according to claim 1, characterized in that, The control module is also used for: If the abnormal information pertains to a topic that has not appeared in the user's history, the interaction module is controlled to output the abnormal information. And / or, The interactive module is controlled to output the recommended question text within a preset time period, which is a time period specified by the user.

10. The active question-answering system according to claim 1, characterized in that, The interaction module includes: a data acquisition submodule, a speech-to-text submodule, a text-to-speech submodule, and a speech playback submodule. The data acquisition submodule is used to receive the first voice data; The speech-to-text submodule is used to convert the first speech data to obtain the first text, and send the first text to the control module; The text-to-speech submodule is used to convert the feedback text to obtain second speech data; The voice playback submodule is used to play the second voice data; The text-to-speech submodule is also used to convert the abnormal information to obtain third speech data, and / or to convert the recommended question text to obtain fourth speech data; The voice playback submodule is also used to play the third voice data and / or the fourth voice data.

11. A proactive question-answering method, characterized in that, The interaction module receives the first input, processes the first input, and obtains the first text; The control module sends the first text to the question-and-answer response module to determine the feedback text, and triggers the hardware detection module and the question recommendation module to run based on the first text; The question-and-answer response module determines the feedback text based on the first text; The hardware detection module collects hardware sensing data, performs abnormal data detection based on the hardware sensing data, generates abnormal information when abnormal data is detected, and sends the abnormal information to the control module. The question recommendation module generates recommended question text based on the first text and sends the recommended question text to the control module; The control module controls the interaction module to output the feedback text, as well as the error information and / or recommended question text; The interaction module outputs the feedback text, as well as the error information and / or recommended question text.

12. A refrigeration device, characterized in that, include: The active question-answering system as described in any one of claims 1 to 10.

Images