Recommendation content generation method and device, equipment and storage medium

Abstract

The embodiment of the invention relates to the technical field of vehicle control, and discloses a recommended content generation method, device and equipment and a storage medium, and the method comprises the steps: splicing an input text of a user with a preset guidance classification text to form a composite text fusing a task instruction and to-be-classified data, the preset guidance classification text is used for guiding the model to map the input text into a specified category set; the composite text is input into a pre-trained large language model for natural language understanding, a classification result is obtained, and the classification result comprises a target category to which the input text belongs in a specified category set; based on the classification result and a pre-established mapping relationship between the category and the vehicle atomization service, generating recommendation content of the next round of dialogue; the recommendation content comprises the target vehicle atomization service corresponding to the target category. By applying the technical scheme of the invention, the problem that the recommended content does not accord with the current demand of the user or the environment in the prior art can be solved.

Description

Technical Field

[0001] Embodiments of the present invention relate to the field of vehicle control technology, and more particularly to a method, apparatus, device, and storage medium for generating recommended content. Background Art

[0002] When a user engages in a conversation with the in-vehicle system, the system responds and makes recommendations based on the user's input and the context of the conversation. These recommendations aim to predict topics the user might want to discuss in the upcoming conversation. However, because these recommendations may not fully consider the user's current context, they may not be suitable for the user's current needs or circumstances. Summary of the Invention

[0003] In view of the above problems, embodiments of the present invention provide a method, apparatus, device, and storage medium for generating recommended content, which are used to solve the problem in the prior art that the recommended content does not meet the user's current needs or environment.

[0004] According to one aspect of an embodiment of the present invention, a method for generating recommended content is provided, the method comprising:

[0005] The user's input text is combined with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to a specified category set;

[0006] Inputting the composite text into a pre-trained large language model for natural language understanding to obtain a classification result, wherein the classification result includes a target category to which the input text belongs in the specified category set;

[0007] Based on the classification result and the pre-established mapping relationship between the category and the vehicle atomic service, the recommended content for the next round of dialogue is generated; the recommended content includes the target vehicle atomic service corresponding to the target category.

[0008] According to another aspect of an embodiment of the present invention, there is provided a device for generating recommended content, comprising:

[0009] A splicing module is used to splice the user's input text with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to a specified category set;

[0010] An input model is used to input the composite text into a pre-trained large language model for natural language understanding, and obtain a classification result, wherein the classification result includes a target category to which the input text belongs in the specified category set;

[0011] A generation module is used to generate recommended content for the next round of dialogue based on the classification results and the pre-established mapping relationship between categories and vehicle atomic services; the recommended content includes the target vehicle atomic service corresponding to the target category.

[0012] According to another aspect of an embodiment of the present invention, there is provided a recommended content generating device, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus;

[0013] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform the operations of the recommended content generation method as described above.

[0014] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores at least one executable instruction, and the executable instruction enables a recommended content generating device / apparatus to perform the operations of the recommended content generating method described above.

[0015] The embodiment of the present invention forms a composite text by splicing the user's input text with the preset guidance classification text. This process ensures that the specific needs of the user input are fully considered. The preset guidance classification text guides the large language model to accurately map the input text to the specified category set. Using a pre-trained large language model for natural language understanding can better understand the user's intentions and context. The target category to which the user input belongs can be accurately identified. This precise matching mechanism ensures that the recommended content is highly relevant to the user's actual needs. Based on the classification results, the mapping relationship between the pre-established categories and the vehicle atomic services is queried. This mapping relationship can flexibly match user needs with the most relevant vehicle atomic services, ensuring the relevance and practicality of the recommended content. In this way, the problem in the prior art that the recommended content does not meet the user's current needs or environment can be solved.

[0016] The above description is only an overview of the technical solutions of the embodiments of the present invention. In order to more clearly understand the technical means of the embodiments of the present invention, they can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiments of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the drawings. In the drawings:

[0018] Figure 1 A schematic diagram showing a flow chart of a first embodiment of a method for generating recommended content provided by the present invention;

[0019] Figure 2 A schematic diagram showing a flow chart of a second embodiment of the method for generating recommended content provided by the present invention;

[0020] Figure 3 A schematic diagram showing the mapping relationship between categories and corpus texts provided by the present invention;

[0021] Figure 4 A schematic diagram showing the mapping relationship between categories and vehicle atomic services provided by the present invention;

[0022] Figure 5 A schematic diagram showing the mapping relationship between the vehicle atomization service provided by the present invention and the corpus text;

[0023] Figure 6 A schematic diagram of a dialogue interface provided by the present invention is shown;

[0024] Figure 7 A schematic diagram showing the execution process of the method for generating recommended content provided by the present invention is shown;

[0025] Figure 8 A schematic structural diagram of a first embodiment of a device for generating recommended content provided by the present invention is shown;

[0026] Figure 9 A schematic structural diagram of an embodiment of a device for generating recommended content provided by the present invention is shown;

[0027] Figure 10 A schematic structural diagram of an embodiment of a vehicle provided by the present invention is shown. DETAILED DESCRIPTION

[0028] The exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0029] Figure 1 FIG1 shows a flow chart of a first embodiment of a method for generating recommended content according to the present invention, which can be executed by a cockpit domain controller. Figure 1 As shown, the method includes the following steps:

[0030] Step 110: The user's input text is concatenated with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to a specified category set.

[0031] The Cockpit Domain Controller (CDC) is an integrated electronic control unit that centrally manages and controls intelligent functions within the vehicle's cockpit. When a user issues a function request (such as adjusting the air conditioning temperature, switching navigation maps, or playing multimedia content) through voice commands, a touch screen, physical buttons, or other interactive methods, the signal is transmitted to the CDC. The CDC processes the signal and sends corresponding instructions to the relevant execution modules in the vehicle, such as the air conditioning system, instrument display, ambient lighting, and seat control units. By coordinating the collaborative operation of these modules, the CDC enables multi-functional integrated control and intelligent interaction within the cockpit, thereby enhancing the comfort, convenience, and technological feel of the cockpit and providing users with an immersive smart cockpit experience.

[0032] In an embodiment of the present invention, the user interacts with the artificial intelligence through the user interface in the vehicle cockpit. This interaction is usually carried out in the form of voice or text. When the user enters text, the information will be transmitted to the cockpit domain controller. The task of the cockpit domain controller is to process these input texts in order to better understand the user's needs. The cockpit domain controller can splice the user's input text with the preset guidance classification text. The preset guidance classification text provides clear classification rules or examples (i.e., "instructions"), which can guide the model to determine whether the input text belongs to a category in a specified category set. By combining the user input text with the guidance classification text, a composite text containing task instructions and data to be classified is formed. The role of this composite text is to provide a clearer context for the large language model so that the model can accurately map the user's input text to a predefined category set.

[0033] Step 120: Input the composite text into a pre-trained large language model for natural language understanding to obtain a classification result, which includes the target category to which the input text belongs in the specified category set.

[0034] The composite text is input into the large language model. The large language model analyzes and understands the composite text. It leverages the language patterns and knowledge acquired during training to parse the text's semantics and intent. Based on its understanding of the text, the model compares the input text against a predefined set of categories. By identifying key features and patterns in the text, the model determines which category the input text best fits. Ultimately, the model outputs a classification result, indicating the target category to which the input text belongs within the specified set of categories.

[0035] Step 130: Based on the classification results and the pre-established mapping relationship between categories and vehicle atomic services, generate recommended content for the next round of dialogue; the recommended content includes the target vehicle atomic service corresponding to the target category.

[0036] The recommended content for the next round of conversation refers to the predicted content that the user may ask in the next round.

[0037] In this embodiment of the present invention, based on the classification results and a pre-established mapping between categories and vehicle atomic services, one or more vehicle atomic services corresponding to the target category are identified. These services are the functions most likely to be of interest or required by the user based on their current needs. Based on the identified target vehicle atomic services, recommended content is generated for the next round of conversation.

[0038] The recommended content generation method provided by the embodiment of the present invention forms a composite text by splicing the user's input text with the preset guidance classification text. This process ensures that the specific needs of the user input are fully considered. The preset guidance classification text guides the large language model to accurately map the input text to the specified category set. Using a pre-trained large language model for natural language understanding can better understand the user's intentions and context. The target category to which the user input belongs can be accurately identified. This precise matching mechanism ensures that the recommended content is highly relevant to the user's actual needs. Based on the classification results, the mapping relationship between the pre-established categories and the vehicle atomic services is queried. This mapping relationship can flexibly match user needs with the most relevant vehicle atomic services, ensuring the relevance and practicality of the recommended content. In this way, the problem in the prior art that the recommended content does not meet the user's current needs or environment can be solved.

[0039] Figure 2 FIG. 1 is a flow chart showing another embodiment of a method for generating recommended content according to the present invention, which can be executed by a cockpit domain controller. Figure 2 As shown, the method includes the following steps:

[0040] Step 210: The user's input text is concatenated with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to a specified category set.

[0041] The predefined guidance classification text can be as follows:

[0042] You are a natural language understanding expert. You need to determine whether the user's input text belongs to one of the following categories: "Thermal Comfort Perception", "Cold Comfort Perception", "Air Quality Perception", "Ventilation Perception", "Optical Component Visibility Perception", "Music Perception", "Itinerary Planning Perception", or "Other". The requirements are as follows:

[0043] If it belongs to any of these categories, just give the category name. If it does not belong to these categories, output "Other". Do not give additional content.

[0044] Output must be one of the following: ['thermal comfort perception', 'cooling comfort perception', 'air quality perception', 'ventilation perception', 'optics visibility perception', 'music perception', 'trip planning perception', or 'other'].

[0045] The user's input text is: ××××××.

[0046] It can be understood that through this guided classification text, the model output result can be guaranteed to be one of the following categories: "thermal comfort perception", "cold comfort perception", "air quality perception", "ventilation perception", "optical component visibility perception", "music perception", "trip planning perception" or "other".

[0047] In an optional manner, the categories in the category set include one or more of thermal comfort perception, cold comfort perception, ventilation perception, air quality perception, and optical component visibility perception.

[0048] Thermal comfort perception involves the perception and description of whether the vehicle's interior temperature is too high or too hot. Cold comfort perception involves the perception and description of whether the vehicle's interior temperature is too low or too cold. Ventilation perception involves the perception and description of the ventilation conditions inside the vehicle. Air quality perception involves the perception and description of the air quality inside the vehicle. Optical component visibility perception involves the perception and description of the visibility and defogger function of optical components such as windows, windshields, and rearview mirrors.

[0049] Reference Figure 3 The figure shows a schematic diagram of the mapping relationship between categories and corpus texts provided by the present invention. The figure displays five different categories: thermal comfort perception, cold comfort perception, ventilation perception, air quality perception, and optical component visibility perception. Each category has several corresponding corpus examples. Not only are Chinese corpus texts included, but also corresponding English translations are provided. When the user's input text matches one of the corpus texts, it can be accurately identified and determined to which category it belongs.

[0050] Step 220: Input the composite text into a pre-trained large language model for natural language understanding to obtain a classification result, which includes the target category to which the input text belongs in the specified category set.

[0051] In an optional method, the composite text is input into a pre-trained large language model for natural language understanding to obtain a classification result. Specifically, the following steps may be included:

[0052] The composite text is input into a pre-trained instruction-fine-tuned large language model, which uses its natural language understanding capabilities to parse task instructions and data content, and generates classification results based on pre-trained knowledge.

[0053] Composite text consists of user input text and preset guidance classification text.

[0054] An instruction-fine-tuned large language model refers to a language model that is based on a large-scale pre-trained language model and is further fine-tuned on a specific instruction dataset, enabling the model to better understand and follow the natural language instructions given by humans and generate output that is more in line with the instruction requirements.

[0055] In this implementation, the composite text can be input into a large, instruction-based fine-tuned language model. The model uses its natural language understanding capabilities to parse the task instructions and data content within the composite text and generates classification results based on the knowledge accumulated during the pre-training phase.

[0056] Step 230: Based on the classification result, query the pre-established mapping relationship between the category and the vehicle atomic service to obtain at least one vehicle atomic service corresponding to the target category.

[0057] Atomic services refer to breaking down complex functions into smaller, more specific services, which allows for more flexible combination and control of these services to meet different needs.

[0058] Reference Figure 4 As shown, a schematic diagram of the mapping relationship between the categories provided by the present invention and the vehicle atomic services is shown. The figure shows that there is a mapping relationship between thermal comfort perception and vehicle defog service and air-conditioning control service, and there is a mapping relationship between air quality perception and window control service and fragrance control service.

[0059] In this embodiment, the database stores a mapping relationship between each category and the vehicle atomic service. According to the identified target category, at least one vehicle atomic service associated with the category can be searched in the database.

[0060] In an optional manner, the vehicle atomic service includes one or more of air conditioning control service, external circulation ventilation service, air purification service, seat heating service, vehicle defog service, window control service, and fragrance control service.

[0061] Step 240: Select a target vehicle atomic service from at least one vehicle atomic service.

[0062] In an embodiment of the present invention, a target vehicle atomic service that best suits the current situation and user needs may be selected from a plurality of possible vehicle atomic services based on the vehicle status information.

[0063] Step 250: Generate recommendation text for the target vehicle atomic service, and / or generate control instructions for activating the target vehicle atomic service as recommended content for the next round of dialogue; the recommended content includes the target vehicle atomic service corresponding to the target category.

[0064] In an optional manner, generating a recommendation text for the atomic service of the target vehicle may specifically include the following steps:

[0065] Based on the target vehicle atomization service, query the pre-established mapping relationship between the vehicle atomization service and the corpus text to obtain at least one corpus text corresponding to the target vehicle atomization service;

[0066] A target corpus text is selected from at least one corpus text to generate a recommendation text for an atomic service of a target vehicle.

[0067] In this implementation, based on the target vehicle atomic service, a pre-established mapping table of vehicle atomic services and corpus texts is queried. This table contains multiple corpus examples corresponding to each service. From the multiple corpus texts retrieved, the most appropriate target corpus text is selected based on specific criteria. These criteria may include current environmental conditions, user preferences, or historical usage data. The selected target corpus text is used to generate the final recommendation text, which is presented to the user on the dialogue interface.

[0068] Reference Figure 5 The figure shows a schematic diagram of the mapping relationship between the vehicle atomic services provided by the present invention and the corpus text. The figure shows five different vehicle atomic services, namely vehicle defog service, air conditioning shutdown service, external circulation ventilation service, air purification service and seat heating service, each of which has several corresponding corpus examples.

[0069] In an optional manner, after generating the recommended content for the next round of conversation, the method for generating recommended content in an embodiment of the present invention may further include the following steps:

[0070] Controls the output of recommended content when answering input text on the interactive interface where the user and artificial intelligence communicate.

[0071] Reference Figure 6 FIG2 is a diagram showing a dialog interface provided by the present invention, on which the answer content and the recommended content are output simultaneously. The recommended content can be presented in the form of text, buttons, links or cards.

[0072] Reference Figure 7As shown, a schematic diagram of the execution process of the recommended content generation method provided by the present invention is shown. When the user inputs text on the dialogue interface, the input text is transmitted to the vehicle's cockpit domain controller. In the cockpit domain controller, the user's input text will be spliced ​​with the preset guidance classification text to form a composite text. Subsequently, the composite text is input into a pre-trained large language model for analysis to obtain a classification result. Based on the classification result, the cockpit domain controller uses the pre-established mapping relationship between the category and the vehicle atomic service to identify one or more vehicle atomic services corresponding to the target category. Finally, the corresponding recommended content is generated and returned to the user. This process not only improves the accuracy of the recommended content, but also enhances the interactive experience between the user and the vehicle system.

[0073] The recommended content generation method provided by the embodiment of the present invention forms a composite text by splicing the user's input text with the preset guidance classification text. This process ensures that the specific needs of the user input are fully considered. The preset guidance classification text guides the large language model to accurately map the input text to the specified category set. Using a pre-trained large language model for natural language understanding can better understand the user's intentions and context. The target category to which the user input belongs can be accurately identified. This precise matching mechanism ensures that the recommended content is highly relevant to the user's actual needs. Based on the classification results, the mapping relationship between the pre-established categories and the vehicle atomic services is queried. This mapping relationship can flexibly match user needs with the most relevant vehicle atomic services, ensuring the relevance and practicality of the recommended content. In this way, the problem in the prior art that the recommended content does not meet the user's current needs or environment can be solved.

[0074] Figure 8 FIG. 1 is a schematic diagram showing the structure of an embodiment of a device for generating recommended content according to the present invention. Figure 8 As shown, the device 800 includes: a splicing module 810, an input module 820 and a generation module 830.

[0075] The splicing module is used to splice the user's input text with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to a specified category set;

[0076] The input model is used to input the composite text into a pre-trained large language model for natural language understanding and obtain a classification result. The classification result includes the target category to which the input text belongs in the specified category set;

[0077] The generation module is used to generate recommended content for the next round of dialogue based on the classification results and the pre-established mapping relationship between categories and vehicle atomic services; the recommended content includes the target vehicle atomic service corresponding to the target category.

[0078] In an optional manner, the generation module is specifically used to:

[0079] Based on the classification result, query the pre-established mapping relationship between the category and the vehicle atomic service to obtain at least one vehicle atomic service corresponding to the target category;

[0080] Selecting a target vehicle atomic service from at least one vehicle atomic service;

[0081] Generate a recommendation text for the target vehicle's atomic service, and / or generate a control instruction for activating the target vehicle's atomic service as recommended content for the next round of dialogue.

[0082] In an optional manner, the input model is specifically used to:

[0083] The composite text is input into a pre-trained instruction-fine-tuned large language model, which uses its natural language understanding capabilities to parse task instructions and data content, and generates classification results based on pre-trained knowledge.

[0084] In an optional manner, the generation module is specifically used to:

[0085] Based on the target vehicle atomization service, query the pre-established mapping relationship between the vehicle atomization service and the corpus text to obtain at least one corpus text corresponding to the target vehicle atomization service;

[0086] A target corpus text is selected from at least one corpus text to generate a recommendation text for an atomic service of a target vehicle.

[0087] In an optional manner, the recommended content generating device is further configured to:

[0088] Controls the output of recommended content when answering input text on the interactive interface where the user and artificial intelligence communicate.

[0089] In an optional manner, the categories in the category set include one or more of thermal comfort perception, cold comfort perception, ventilation perception, air quality perception, and optical component visibility perception.

[0090] In an optional manner, the vehicle atomic service includes one or more of air conditioning control service, external circulation ventilation service, air purification service, seat heating service, and vehicle defog service.

[0091] The recommended content generation method provided by the embodiment of the present invention forms a composite text by splicing the user's input text with the preset guidance classification text. This process ensures that the specific needs of the user input are fully considered. The preset guidance classification text guides the large language model to accurately map the input text to the specified category set. Using a pre-trained large language model for natural language understanding can better understand the user's intentions and context. The target category to which the user input belongs can be accurately identified. This precise matching mechanism ensures that the recommended content is highly relevant to the user's actual needs. Based on the classification results, the mapping relationship between the pre-established categories and the vehicle atomic services is queried. This mapping relationship can flexibly match user needs with the most relevant vehicle atomic services, ensuring the relevance and practicality of the recommended content. In this way, the problem in the prior art that the recommended content does not meet the user's current needs or environment can be solved.

[0092] Figure 9 The schematic diagram shows the structure of an embodiment of the recommended content generation device of the present invention. The specific embodiment of the present invention does not limit the specific implementation of the recommended content generation device.

[0093] like Figure 9 As shown, the recommended content generating device may include: a processor (processor) 902 , a communication interface (Communications Interface) 904 , a memory (memory) 906 , and a communication bus 908 .

[0094] Processor 902, communication interface 904, and memory 906 communicate with each other via communication bus 908. Communication interface 904 is used to communicate with other devices, such as clients or other server network elements. Processor 902 is used to execute program 910, which may specifically perform the steps described in the aforementioned embodiment of the method for generating recommended content.

[0095] Specifically, the program 910 may include program code including computer-executable instructions.

[0096] Processor 902 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included in the recommended content generation device may be of the same type, such as one or more CPUs, or may be of different types, such as one or more CPUs and one or more ASICs.

[0097] The memory 906 is used to store the program 910. The memory 906 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage.

[0098] The program 910 may be specifically called by the processor 902 to enable the recommended content generation device to perform the following operations:

[0099] The user's input text is spliced ​​with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to the specified category set;

[0100] Input the composite text into a pre-trained large language model for natural language understanding and obtain a classification result. The classification result includes the target category to which the input text belongs in the specified category set.

[0101] Based on the classification results and the pre-established mapping relationship between categories and vehicle atomic services, recommended content for the next round of conversation is generated; the recommended content includes the target vehicle atomic services corresponding to the target category.

[0102] In an optional manner, the program 910 is called by the processor 902 to enable the recommended content generation device to perform the following operations:

[0103] Based on the classification result, query the pre-established mapping relationship between the category and the vehicle atomic service to obtain at least one vehicle atomic service corresponding to the target category;

[0104] Selecting a target vehicle atomic service from at least one vehicle atomic service;

[0105] Generate a recommendation text for the target vehicle's atomic service, and / or generate a control instruction for activating the target vehicle's atomic service as recommended content for the next round of dialogue.

[0106] In an optional manner, the program 910 is called by the processor 902 to enable the recommended content generation device to perform the following operations:

[0107] The composite text is input into a pre-trained instruction-fine-tuned large language model, which uses its natural language understanding capabilities to parse task instructions and data content, and generates classification results based on pre-trained knowledge.

[0108] In an optional manner, the program 910 is called by the processor 902 to enable the recommended content generation device to perform the following operations:

[0109] Based on the target vehicle atomization service, query the pre-established mapping relationship between the vehicle atomization service and the corpus text to obtain at least one corpus text corresponding to the target vehicle atomization service;

[0110] A target corpus text is selected from at least one corpus text to generate a recommendation text for an atomic service of a target vehicle.

[0111] In an optional manner, the program 910 is called by the processor 902 to enable the recommended content generation device to perform the following operations:

[0112] Controls the output of recommended content when answering input text on the interactive interface where the user and artificial intelligence communicate.

[0113] In an optional manner, the categories in the category set include one or more of thermal comfort perception, cold comfort perception, ventilation perception, air quality perception, and optical component visibility perception.

[0114] In an optional manner, the vehicle atomic service includes one or more of air conditioning control service, external circulation ventilation service, air purification service, seat heating service, and vehicle defog service.

[0115] The recommended content generation method provided by the embodiment of the present invention forms a composite text by splicing the user's input text with the preset guidance classification text. This process ensures that the specific needs of the user input are fully considered. The preset guidance classification text guides the large language model to accurately map the input text to the specified category set. Using a pre-trained large language model for natural language understanding can better understand the user's intentions and context. The target category to which the user input belongs can be accurately identified. This precise matching mechanism ensures that the recommended content is highly relevant to the user's actual needs. Based on the classification results, the mapping relationship between the pre-established categories and the vehicle atomic services is queried. This mapping relationship can flexibly match user needs with the most relevant vehicle atomic services, ensuring the relevance and practicality of the recommended content. In this way, the problem in the prior art that the recommended content does not meet the user's current needs or environment can be solved.

[0116] Figure 10 FIG. 1 shows a schematic structural diagram of an embodiment of a vehicle of the present invention. Figure 10 As shown, the vehicle 1000 includes: sensors, cameras, one or more processors and a communication interface;

[0117] Sensors and cameras are used to obtain vehicle status information;

[0118] The processor is configured to execute the steps in the above-mentioned embodiment of the method for generating recommended content.

[0119] The recommended content generation method provided by the embodiment of the present invention forms a composite text by splicing the user's input text with the preset guidance classification text. This process ensures that the specific needs of the user input are fully considered. The preset guidance classification text guides the large language model to accurately map the input text to the specified category set. Using a pre-trained large language model for natural language understanding can better understand the user's intentions and context. The target category to which the user input belongs can be accurately identified. This precise matching mechanism ensures that the recommended content is highly relevant to the user's actual needs. Based on the classification results, the mapping relationship between the pre-established categories and the vehicle atomic services is queried. This mapping relationship can flexibly match user needs with the most relevant vehicle atomic services, ensuring the relevance and practicality of the recommended content. In this way, the problem in the prior art that the recommended content does not meet the user's current needs or environment can be solved.

[0120] An embodiment of the present invention provides a computer-readable storage medium storing at least one executable instruction. When the executable instruction runs on a recommended content generation device / apparatus, the recommended content generation device / apparatus executes the recommended content generation method in any of the above method embodiments.

[0121] The executable instructions may be specifically used to enable the recommended content generating device / apparatus to perform the following operations:

[0122] The user's input text is spliced ​​with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to the specified category set;

[0123] Input the composite text into a pre-trained large language model for natural language understanding and obtain a classification result. The classification result includes the target category to which the input text belongs in the specified category set.

[0124] Based on the classification results and the pre-established mapping relationship between categories and vehicle atomic services, recommended content for the next round of conversation is generated; the recommended content includes the target vehicle atomic services corresponding to the target category.

[0125] In an optional manner, the executable instructions enable the recommended content generating device / apparatus to perform the following operations:

[0126] Based on the classification result, query the pre-established mapping relationship between the category and the vehicle atomic service to obtain at least one vehicle atomic service corresponding to the target category;

[0127] Selecting a target vehicle atomic service from at least one vehicle atomic service;

[0128] Generate a recommendation text for the target vehicle's atomic service, and / or generate a control instruction for activating the target vehicle's atomic service as recommended content for the next round of dialogue.

[0129] In an optional manner, the executable instructions enable the recommended content generating device / apparatus to perform the following operations:

[0130] The composite text is input into a pre-trained instruction-fine-tuned large language model, which uses its natural language understanding capabilities to parse task instructions and data content, and generates classification results based on pre-trained knowledge.

[0131] In an optional manner, the executable instructions enable the recommended content generating device / apparatus to perform the following operations:

[0132] Based on the target vehicle atomization service, query the pre-established mapping relationship between the vehicle atomization service and the corpus text to obtain at least one corpus text corresponding to the target vehicle atomization service;

[0133] A target corpus text is selected from at least one corpus text to generate a recommendation text for an atomic service of a target vehicle.

[0134] In an optional manner, the executable instructions enable the recommended content generating device / apparatus to perform the following operations:

[0135] Controls the output of recommended content when answering input text on the interactive interface where the user and artificial intelligence communicate.

[0136] In an optional manner, the categories in the category set include one or more of thermal comfort perception, cold comfort perception, ventilation perception, air quality perception, and optical component visibility perception.

[0137] In an optional manner, the vehicle atomic service includes one or more of air conditioning control service, external circulation ventilation service, air purification service, seat heating service, and vehicle defog service.

[0138] The recommended content generation method provided by the embodiment of the present invention forms a composite text by splicing the user's input text with the preset guidance classification text. This process ensures that the specific needs of the user input are fully considered. The preset guidance classification text guides the large language model to accurately map the input text to the specified category set. Using a pre-trained large language model for natural language understanding can better understand the user's intentions and context. The target category to which the user input belongs can be accurately identified. This precise matching mechanism ensures that the recommended content is highly relevant to the user's actual needs. Based on the classification results, the mapping relationship between the pre-established categories and the vehicle atomic services is queried. This mapping relationship can flexibly match user needs with the most relevant vehicle atomic services, ensuring the relevance and practicality of the recommended content. In this way, the problem in the prior art that the recommended content does not meet the user's current needs or environment can be solved.

[0139] The algorithms or displays provided herein are not inherently related to any particular computer, virtual system or other device. In addition, the embodiments of the present invention are not directed to any particular programming language.

[0140] In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the present invention may be practiced without these specific details. Similarly, in order to streamline the present invention and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. The claims that follow the detailed description are hereby expressly incorporated into that detailed description, with each claim itself serving as a separate embodiment of the present invention.

[0141] Those skilled in the art will appreciate that the modules in the devices of the embodiments can be adaptively changed and installed in one or more devices different from the embodiments. The modules, units, or components in the embodiments can be combined into one module, unit, or component, and furthermore, they can be divided into multiple submodules, subunits, or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive.

[0142] It should be noted that the above embodiments illustrate rather than limit the invention, and that alternative embodiments may be devised by a person skilled in the art without departing from the scope of the appended claims. In the claims, any reference signs placed between brackets should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising several different elements and by means of appropriately programmed computers. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments should not be understood as limiting the order of execution unless otherwise specified.

Claims

1. A method for generating recommended content, characterized in that: The method comprises: The user's input text is combined with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to a specified category set; Inputting the composite text into a pre-trained large language model for natural language understanding to obtain a classification result, wherein the classification result includes a target category to which the input text belongs in the specified category set; Based on the classification result and the pre-established mapping relationship between the category and the vehicle atomic service, the recommended content for the next round of dialogue is generated; the recommended content includes the target vehicle atomic service corresponding to the target category.

2. The method according to claim 1, characterized in that The generating of recommended content for the next round of conversation based on the classification results and the pre-established mapping relationship between the categories and the vehicle atomic services includes: Based on the classification result, querying the pre-established mapping relationship between the category and the vehicle atomic service to obtain at least one vehicle atomic service corresponding to the target category; selecting a target vehicle atomic service from the at least one vehicle atomic service; Generate a recommendation text for the target vehicle atomic service, and / or generate a control instruction for activating the target vehicle atomic service as recommended content for the next round of dialogue.

3. The method according to claim 1, characterized in that Inputting the composite text into a pre-trained large language model for natural language understanding to obtain a classification result includes: The composite text is input into a pre-trained instruction fine-tuning large language model, which uses its natural language understanding capability to parse the task instructions and data content, and generates classification results based on the pre-trained knowledge.

4. The method according to claim 2, characterized in that The generating of the recommendation text for the atomic service of the target vehicle includes: Based on the target vehicle atomization service, querying a pre-established mapping relationship between the vehicle atomization service and the corpus text to obtain at least one corpus text corresponding to the target vehicle atomization service; A target corpus text is selected from the at least one corpus text to generate a recommendation text for the atomic service of the target vehicle.

5. The method according to any one of claims 1 to 4, characterized in that After generating the recommended content for the next round of dialogue, the method further includes: The recommended content is output when the answer content of the input text is output on the interactive interface of the user's dialogue with the artificial intelligence.

6. The method according to any one of claims 1 to 4, characterized in that The categories in the category set include one or more of thermal comfort perception, cold comfort perception, ventilation perception, air quality perception, and optical component visibility perception.

7. The method according to any one of claims 1 to 4, characterized in that The vehicle atomic services include one or more of air conditioning control services, external circulation ventilation services, air purification services, seat heating services, and vehicle defog services.

8. A recommended content generating device, characterized in that: The device comprises: A splicing module is used to splice the user's input text with the preset guidance classification text to form a composite text that integrates the task instructions and the data to be classified. The preset guidance classification text is used to guide the model to map the input text to a specified category set; An input model is used to input the composite text into a pre-trained large language model for natural language understanding, and obtain a classification result, wherein the classification result includes a target category to which the input text belongs in the specified category set; A generation module is used to generate recommended content for the next round of dialogue based on the classification results and the pre-established mapping relationship between categories and vehicle atomic services; the recommended content includes the target vehicle atomic service corresponding to the target category.

9. A recommended content generating device, characterized in that: include: A processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform the operation of the recommended content generation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The storage medium stores at least one executable instruction. When the executable instruction is executed on the recommended content generating device / apparatus, the recommended content generating device / apparatus performs the operation of the recommended content generating method according to any one of claims 1 to 7.

Images