Interaction method and device, electronic equipment and storage medium

Abstract

The invention discloses an interaction method and device, electronic equipment and a storage medium. The method comprises the steps that a first model is obtained through training of vehicle data corresponding to a vehicle which supports diagnosis of a vehicle diagnosis product, then an interaction device responds to input operation of a user and obtains a first problem, and at the moment, the first problem is used for querying vehicle information which supports diagnosis; obtaining a first answer corresponding to the first question based on the first question and a first model; then the interaction device displays the first answer, that is, the user is helped to inquire or determine the vehicle information supporting diagnosis through an artificial intelligence mode, the user does not need to manually rummage and inquire on an interface corresponding to a vehicle diagnosis product, efficiency and accuracy are achieved, and the user experience can be improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and particularly to an interaction method, device, electronic device, and storage medium. Background Art

[0002] With the development of artificial intelligence technology and the rise of large language models, the automotive industry has increasingly used this technology to meet user needs.

[0003] When a user wants to determine the vehicle information supported by a certain vehicle diagnostic product, such as determining vehicle information such as the supported vehicle brand, model, year, function, etc., currently the user mainly determines it by manually searching and querying on the interface corresponding to the vehicle diagnostic product (this interface includes the vehicle information supported for diagnosis). Due to the huge amount of vehicle information data such as the supported vehicle brands and models of the vehicle diagnostic product, problems such as a complicated query process and incomplete information display will occur during the user's query of the web page, which not only results in a low determination accuracy but also low efficiency, and cannot meet the user's need to simply and efficiently obtain the vehicle information supported for diagnosis.

[0004] Therefore, how to efficiently and accurately determine the vehicle information supported for diagnosis to improve the user experience is an urgent problem to be solved. Summary of the Invention

[0005] This application provides an interaction method, device, electronic device, and storage medium, which can efficiently and accurately determine the vehicle information supported for diagnosis and improve the user experience.

[0006] In a first aspect, this application provides an interaction method, which is applied to an interaction device, and the method includes:

[0007] In response to a user's input operation, obtain a first question, where the first question is used to query the vehicle information supported for diagnosis;

[0008] Display a first answer corresponding to the first question, where the first answer is obtained based on the first question and a first model, and the first model is trained through vehicle data corresponding to the vehicles supported for diagnosis by the vehicle diagnostic product.

[0009] In a second aspect, this application provides an interaction method, which is applied to a server, and the method includes:

[0010] Obtain a first question, where the first question is used to query the vehicle information supported for diagnosis;

[0011] Based on the first question and the first model, generate a first answer corresponding to the first question, where the first model is trained through vehicle data corresponding to the vehicles supported for diagnosis by the vehicle diagnostic product.

[0012] In a third aspect, the present application provides an interaction device, which includes: a first acquisition unit and a first processing unit;

[0013] The first acquisition unit is configured to acquire a first question in response to a user's input operation, where the first question is used to query vehicle information supported for diagnosis;

[0014] The first processing unit is configured to display a first answer corresponding to the first question, where the first answer is obtained based on the first question and a first model, and the first model is trained by vehicle data corresponding to vehicles supported for diagnosis by a vehicle diagnosis product.

[0015] In a fourth aspect, the present application provides a server, which includes: a second acquisition unit and a second processing unit;

[0016] The second acquisition unit is configured to acquire a first question, where the first question is used to query vehicle information supported for diagnosis;

[0017] The second processing unit is configured to generate a first answer corresponding to the first question based on the first question and a first model, and the first model is trained by vehicle data corresponding to vehicles supported for diagnosis by a vehicle diagnosis product.

[0018] In a fifth aspect, the present application provides an electronic device, which includes: a processor and a memory. The processor is connected to the memory, and the memory is used to store a computer program. The processor is configured to execute the computer program stored in the memory so that the electronic device executes the methods in the first aspect and the second aspect.

[0019] In a sixth aspect, the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the methods in the first aspect and the second aspect are executed.

[0020] In a seventh aspect, the present application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the methods in the first aspect and the second aspect are executed.

[0021] Implementing the present application has the following beneficial effects:

[0022] In an embodiment of the present application, a first model is trained using vehicle data corresponding to vehicles supported by a vehicle diagnostic product. Then, in response to a user's input operation, an interaction device obtains a first question, where the first question is used to query vehicle information supported by the diagnosis. Then, based on the first question and the first model, a first answer corresponding to the first question is obtained. Then, the interaction device displays the first answer, that is, in an artificial intelligence manner, it helps the user query or determine the vehicle information supported by the diagnosis, without the user manually searching through the interface corresponding to the vehicle diagnostic product, which is not only efficient and accurate but also can enhance the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these drawings without creative efforts.

[0024] Figure 1 Structural schematic diagram of a first model provided by an embodiment of the present application;

[0025] Figure 2 Schematic diagram of an interaction system provided by an embodiment of the present application;

[0026] Figure 3 Structural schematic diagram of a knowledge graph based on vehicle data supported by diagnosis provided by an embodiment of the present application;

[0027] Figure 4 Flow schematic diagram of an interaction method provided by an embodiment of the present application;

[0028] Figure 5 Schematic diagram for determining user intent based on a preset function button provided by an embodiment of the present application;

[0029] Figure 6 Interaction flow schematic diagram of an interaction method provided by an embodiment of the present application;

[0030] Figure 7 Training method of a first model provided by an embodiment of the present application;

[0031] Figure 8 Another structural schematic diagram of a first model provided by an embodiment of the present application;

[0032] Figure 9 Still another structural schematic diagram of a first model provided by an embodiment of the present application;

[0033] Figure 10 Scene schematic diagram provided by an embodiment of the present application;

[0034] Figure 11 Another schematic diagram of a scenario provided by an embodiment of the present application;

[0035] Figure 12 A block diagram of the functional units of an interaction device provided by an embodiment of the present application;

[0036] Figure 13 A block diagram of the functional units of a server provided by an embodiment of the present application;

[0037] Figure 14 A schematic structural diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0038] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0039] The terms "first", "second", "third", and "fourth" in the specification, claims, and drawings of the present application are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products, or devices.

[0040] Referring to "embodiment" in this context means that a particular feature, result, or characteristic described in connection with the embodiment can be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0041] First, relevant terms and related technologies involved in the embodiments of the present application are explained:

[0042] First Model: In the embodiments of the present application, the first model may be a large model based on the Transformer architecture, such as the GPT (Generative Pre-trained Transformer) model. The first model may mainly consist of multiple Decoders in the Transformer architecture.

[0043] Exemplarily, refer to Figure 1 , Figure 1 which is a schematic structural diagram of a first model provided by an embodiment of the present application.

[0044] As Figure 1 shown, the first model includes N Decoders. Each decoder mainly includes an Embedding layer, a Masked Multi-head Attention layer, a Norm layer, a Feed-forward Neural Network, a Linear layer, and a Softmax activation layer. Taking one decoder as an example, for the input Input of the first model, the input Input is vector-embedded by Embedding and added with Positional Encoding to obtain a first feature vector corresponding to the input. Then, the first feature vector is processed by the masked multi-head attention mechanism to obtain a second feature vector corresponding to the input. Then, the first feature vector and the second feature vector corresponding to the input are subjected to residual connection Add and then normalized by Norm to obtain a third feature vector corresponding to the input. Then, the third feature vector is input into the feed-forward neural network for processing to obtain a fourth feature vector corresponding to the input. Then, the third feature vector and the fourth feature vector are subjected to residual connection Add and then normalized by Norm to obtain a fifth feature vector corresponding to the input. Then, the fifth feature vector is input into the Linear layer for linear processing and then activation processing to obtain a predicted probability corresponding to the input, generally a probability distribution corresponding to the input. Furthermore, the final output Output can be determined based on the predicted probability. Specifically:

[0045] If it is a discriminative classification problem, such as binary classification, multi-classification, etc., then the predicted probability obtained based on the Softmax processing can be understood as the predicted probability of the input under each classification category, that is, a probability distribution. If it belongs to the classification category, the probability is 1, otherwise it is 0. Its corresponding label is the true probability of the input under each classification category. Furthermore, the classification category with the highest probability in the probability distribution corresponding to the input can be determined as the output corresponding to the input.

[0046] If it is a generative problem, such as taking a question as input and generating a corresponding answer as output. Since generative methods are autoregressive, unlike discriminative methods that can directly obtain a complete output through a probability distribution processed by Linear and Softmax, that is, it does not output a complete output sequence at once, but gradually predicts and generates elements in the output sequence until a complete output is obtained.

[0047] First, refer to Figure 2 , Figure 2 which is a schematic diagram of an interaction system provided by an embodiment of the present application.

[0048] As Figure 2 shown, Figure 2 the interaction system shown includes an interaction device and a server. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, and big data and artificial intelligence platforms. The present application does not make specific limitations. The interaction device can be a user terminal, such as a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart TV, a desktop computer, a smart watch, a smart vehicle, etc., but is not limited thereto; the number of user terminals can be one or more, and the present application also does not make limitations; a target application or a target web page can be installed on the interaction device, and the user can perform data query interactions (such as querying vehicle information supported for diagnosis, etc.) through the target application or the target web page to implement the human-computer interaction function. Specifically:

[0049] The user can perform data query operations through the target application or the target web page on the interaction device. For example, the user can perform input operations on the first interface of the target application or the target web page, and then the interaction device responds to the user's input operation and obtains the first question input by the user, where the first question is used to represent the vehicle information supported for diagnosis that the user wants to query; then the interaction device can send the first question to the server; at this time, a first model is deployed on the server, and the first model is trained by vehicle data corresponding to the vehicles supported by the vehicle diagnosis product (it should be noted that the training method of the first model is not elaborated here, and specific reference can be made to the corresponding explanations in the following embodiments), and then the server inputs the first question into the first model and outputs the first answer corresponding to the first question; then the server sends the first answer to the interaction device, and correspondingly, the interaction device receives the first answer and displays the first answer. At this time, the first question and the first answer form a conversation (also called a round of conversation). By analogy, the user can perform multiple rounds of conversations through the interaction device, and the principle of each round of conversation is similar.

[0050] It should be noted that Figure 2 The interaction device and the server in the interaction system shown can also execute other corresponding steps in the following embodiments, which will not be elaborated one by one here. For specific reference, please refer to the following embodiments.

[0051] It should be noted that based on the interaction method of the present application, the user can query corresponding data, such as querying vehicle information supported for diagnosis, querying vehicle maintenance plans, vehicle sales situations, and any other questions the user wants to query. The present application does not make any limitations. The following embodiments of the present application mainly take the intention of querying vehicle information supported for diagnosis as an example for explanation. Before introducing the method embodiments, first, the vehicle data corresponding to the vehicles supported by the vehicle diagnosis products involved in the embodiments of the present application will be explained as follows:

[0052] In the embodiments of the present application, the vehicle data corresponding to the vehicles supported by the vehicle diagnosis products can be collected, which mainly includes module contents such as vehicle brands supported for diagnosis, vehicle models, and vehicle years. In addition, it can also include module contents such as vehicle systems and vehicle functions, which will not be listed one by one in the present application; vehicle types can include various types of vehicles such as commercial vehicles, passenger vehicles, motorcycles, and new energy vehicles, which will not be listed one by one here either. Under each type of vehicle, there can be different brands of vehicles, and each brand of vehicle can include different models, years, systems, functions, etc.

[0053] Furthermore, a corresponding first database can be constructed based on these vehicle data supported by the vehicle diagnosis products. At this time, each piece of data in this first database can be stored in a first preset format. Through all the data in the first database, the corresponding relationship or the association relationship between the various module contents in the above vehicle data can be completely reflected, such as all vehicle models under a brand.

[0054] Taking the module contents included in the above vehicle data, namely vehicle brand, vehicle model, vehicle year, vehicle system, and vehicle function, as an example for explanation, for example, the first preset format is "vehicle type_vehicle brand_vehicle model_vehicle year_vehicle system_vehicle function". Among them, the contents of each module in the first preset format can be exchanged in position, and the number of contents belonging to the same module can be one or more, such as there are multiple vehicle models and multiple vehicle years; "_" represents a separator (other symbols can also be used as separators, which is not limited in the present application); for some vehicles, if some of the contents such as vehicle type, vehicle brand, vehicle model, vehicle year, vehicle system, and vehicle function do not exist, the corresponding parts in the data will be NULL, and finally, any two pieces of data are different. For example:

[0055] First data: Brand 1_Model 1_Year 1_System 1_Function 1;

[0056] Second data: Brand 1_Model 2_Year 2_System 1_Function 1;

[0057] Third data: Brand 1_Model 3_Year 3_System 1_NULL.

[0058] In an alternative embodiment, the first database may also be a knowledge graph generated based on vehicle data. In this case, template contents such as vehicle type, vehicle brand, vehicle signal, vehicle year, vehicle system, and vehicle function in the vehicle data can be used as entities in the knowledge graph, and the corresponding relationships or associations between the template contents such as vehicle type, vehicle brand, vehicle signal, vehicle year, vehicle system, and vehicle function are represented by connection lines, and a knowledge graph corresponding to the vehicle data supporting diagnosis is generated together.

[0059] For example, refer to Figure 3 , Figure 3 which is a schematic structural diagram of a knowledge graph based on vehicle data supporting diagnosis provided by an embodiment of the present application. As Figure 3 shown, Figure 3 each content in

[0060] is used as an entity in the knowledge graph, and the connection lines indicate the existence of an association relationship or a corresponding relationship. Specifically: The models corresponding to Brand 1 include Model 1-1, Model 1-2, Model 1-3, Model 1-4, Model 1-5, and Model 1-6; The years corresponding to Brand 1 and Model 1-1 include Year 2, the years corresponding to Brand 1 and Model 1-2 include Year 2, the years corresponding to Brand 1 and Model 1-3 include Year 1, the years corresponding to Brand 1 and Model 1-4 include Year 3, and the corresponding system includes System 2, the years corresponding to Brand 1 and Model 1-5 include Year 3, and the corresponding system includes System 2, the years corresponding to Brand 1 and Model 1-6 include Year 4, and the corresponding system includes System 1. Similarly, the models corresponding to Brand 2 include Model 2-1, Model 2-2, Model 2-3, and Model 2-4; The years corresponding to Brand 2 and Model 2-1 include Year 5, the years corresponding to Brand 2 and Model 2-2 include Year 1, the years corresponding to Brand 2 and Model 2-3 include Year 3, and the corresponding system includes System 2, the years corresponding to Brand 2 and Model 2-4 include Year 5, and the corresponding system includes System 3, and the corresponding function includes Function 1. Figure 4 , Figure 4 which is a schematic flowchart of an interaction method provided by an embodiment of the present application. This method is applied to the interaction device in the above embodiment; This method includes but is not limited to steps S401-S402:

[0061] S401. The interaction device responds to the user's input operation to obtain a first question, which is used to query vehicle information supported by diagnosis.

[0062] In an embodiment of the present application, the user can input the first question through the interaction device (such as the target application or the target web page, which will not be elaborated further); then the interaction device responds to the user's input operation to obtain the first question input by the user, where the first question can be used to represent querying vehicle information supported by diagnosis; then the interaction device sends the first question to the server.

[0063] S402. The interaction device displays a first answer corresponding to the first question, where the first answer is obtained based on the first question and the first model, and the first model is trained by vehicle data corresponding to vehicles supported by vehicle diagnostic products.

[0064] Correspondingly, the server receives the first question from the interaction device, and then the server generates a first answer corresponding to the first question based on the first question and the first model; then the server sends the first answer to the interaction device. Correspondingly, the interaction device receives the first answer and displays it for the user to view. In addition, the form of the first question can be in text form or voice form, which is not limited in this application. When the first question received by the server is in voice form, the server can perform text conversion on the first question, such as based on Automatic Speech Recognition (ASR) technology, and then perform subsequent steps based on the converted text, which will not be elaborated further here.

[0065] Among them, the server generates a first answer corresponding to the first question based on the first question and the first model, which specifically includes steps S11 - S12:

[0066] S11. Obtain the first intention of the user, which is used to determine vehicle information supported by diagnosis.

[0067] In an embodiment of the present application, the method for obtaining the first intention can at least include the following two methods:

[0068] (1) After the server receives the first question from the interaction device, it performs keyword recognition on the first question to obtain the first keyword(s). The number of the first keywords can be one or more, which is not limited in this application. Then, the server matches the first keyword(s) with the preset keywords corresponding to each preset intention among multiple preset intentions, and determines the first intention from the multiple preset intentions. For example, it calculates the similarity between the first keyword(s) and the preset keywords corresponding to each preset intention. The specific principle for determining the similarity is not limited in this application. Then, the server determines the preset intention corresponding to the preset keyword with the maximum similarity as the first intention. That is to say, in this method, the server does not receive the user's first intention from the interaction device, but determines the first intention based on the first question input by the user.

[0069] (2) The first interface or the target web page of the target application program of the interaction device may include at least one preset function button, and each preset function button corresponds to an intention. For example, the intention can be to determine / query the vehicle information supported by the current vehicle diagnostic product, obtain / query the repair plan for the vehicle, query the after-sales rules corresponding to the sold products, and so on. Then, if the user wants to determine / query the supported vehicle information, the user can touch or select the first function button. Then, the interaction device responds to the touch operation of the user on the first function button and obtains the user's first intention. Then, after the interaction device obtains the first question input by the user, in addition to sending the first question to the server, the interaction device also sends the first intention to the server. Correspondingly, the server obtains the first question and the first intention. That is to say, in this method, the user's intention is that the user selects the first function button corresponding to the first intention through the interaction device, and then the interaction device directly sends the first intention to the server. Of course, if the user does not select the preset function button corresponding to the intention, the intention of the user still needs to be determined based on the first method, which will not be elaborated here.

[0070] For ease of understanding, the following is an explanatory illustration with reference to the accompanying drawings. Refer to Figure 5 , Figure 5 which is a schematic diagram for determining the user intention based on the preset function button provided by the embodiment of this application. As Figure 5 shown, Figure 5The interface shown is the first interface of the target application or the target web page displayed by the interactive device, including an input box (for the user to input content, such as a question), a function button corresponding to Function 1, a first function button corresponding to Function 2, a function button corresponding to Function 3, and a send function button; each of Function 1, Function 2, and Function 3 corresponds to an intention, such as Function 2 representing determining / querying vehicle information supported; the number of function buttons can be multiple, and in this application, only 3 are taken as an example for illustration. For example, when the user wants to determine / query vehicle information supported, the user can select the first function button, and input a first question in the input box, and then click "send". The interactive device obtains the first intention corresponding to the first function button and the first question and sends the first intention and the first question to the server. Additionally, it should be noted that there is no sequence requirement between the user selecting the first function button and inputting the first question in the input box, and this application does not make a limitation. Of course, the user can also not select the corresponding first function button, only input the first question in the input box, and then click "send". The interactive device obtains the first question and sends the first question to the server, and the server determines the first intention based on the first question.

[0071] S12. Obtain a first answer based on the first question, the database corresponding to the first intention, and the first model.

[0072] Exemplarily, first, the server can retrieve from the first database corresponding to the first intention based on the first question to obtain first data related to the first question. The first database is generated based on the vehicle data corresponding to the vehicles supported by the vehicle diagnostic product for diagnosis. In the embodiments of this application, databases corresponding to each intention can be pre-constructed. For example, construct the first database corresponding to the first intention, and the principle will not be elaborated here. Also, for example, for the intention of obtaining / querying the vehicle repair plan, a database corresponding to this intention can be formed based on various repair data of the vehicle. Then the server retrieves from the first database corresponding to the first intention based on the first question to obtain first data related to the first question. For example, by calculating the similarity between the first question and the data in the first database (the specific principle is not limited in this application); then take the data with a similarity greater than the first threshold as the first data.

[0073] Then the server generates a first prompt word based on the first question and the first data. For example, the server can splice the first question and the first data as the first prompt word; or, the server can embed the first question and the first data into a preset prompt word template according to the preset prompt word template to obtain the first prompt word.

[0074] Then, based on the first prompt and the first model, the server obtains the first answer. For example, the server can input the first prompt into the first model, that is, use the first prompt as the input of the first model, and output the first answer corresponding to the first question. At this time, when using the first prompt as the input of the first model, the specific steps executed can refer to the steps executed by the first model on the input Input in the above Figure 1 embodiment, which will not be elaborated here.

[0075] In an alternative embodiment, in terms of obtaining the first answer based on the first question, the database corresponding to the first intent, and the first model, after the server obtains the first intent, it can generate multiple third answers based on the first database and multiple preset reply templates, and the multiple third answers correspond one-to-one with the multiple preset reply templates; then the server determines the first matching result corresponding to the first question from the multiple third answers based on the first question. For example, it can calculate the similarity between the first question and each third answer, and then use the third answer with a similarity greater than the second threshold as the first matching result; then the server generates the first answer based on the first matching result, the first database, the first question, and the first model. Specifically:

[0076] At this time, the first database is a knowledge graph generated based on vehicle data. Then, if the first matching result is empty, the server performs entity annotation on the first question according to the entities in the knowledge graph (not elaborated here), obtaining a fourth question. For example, it can identify the third entity in the first question, and if the knowledge graph includes the third entity, the third entity in the first question is annotated. There are many forms of annotation. For example, the third entity can be annotated with a delimiter, which is not specifically limited in this application. Then the server generates the first answer based on the fourth question, the knowledge graph, and the first model, that is, the server uses the fourth question as the prompt and inputs it into the first model together with the knowledge graph. Then the server executes the following steps through the first model:

[0077] First, feature extraction is performed on the fourth question to obtain the ninth feature, that is, it can be understood as Figure 1 the steps corresponding to the embedding layer and position encoding in

[0078] It should be noted that in this application, a matching feature extractor can be used to extract features from the fourth question and the knowledge graph respectively. For example, a first feature extractor is used to extract features from the fourth question (such as the embedding layer and position encoding corresponding to Figure 1 ), and a second feature extractor is used to extract features from the knowledge graph. The second feature extractor can be an encoder for extracting features from images.

[0079] Then, through graph convolution processing on the second feature, a third feature is obtained. For example, graph convolution processing is performed on the second feature through a Graph Convolutional Networks (GCN). The specific principle of graph convolution is not elaborated here, and the corresponding description in the following embodiments can be specifically referred to. Then, based on the ninth feature and the third feature, attention processing is performed to obtain a tenth feature. For example, cross-attention processing can be performed on the ninth feature and the third feature, or self-attention processing can be performed on the ninth feature and the third feature respectively and then cross-attention processing can be performed. This application does not make a limitation. Finally, based on the tenth feature, a second answer is generated. For example, the fourth feature is processed according to the Figure 1 steps performed by the first model on the first feature vector in the embodiment to process the tenth feature to obtain a first answer. Specifically:

[0080] Masked multi-head attention mechanism processing is performed on the tenth feature to obtain an eleventh feature; then the tenth feature and the eleventh feature vector are subjected to residual connection and then normalization processing to obtain a twelfth feature corresponding to the input; then the twelfth feature is input into a feed-forward neural network for processing to obtain a thirteenth feature; then the twelfth feature and the thirteenth feature are subjected to residual connection and then normalization Norm processing to obtain a fourteenth feature; then the fourteenth feature is input into a linear layer for linear processing and then activation processing to obtain a prediction probability corresponding to the first question, and further a first answer corresponding to the first prediction probability can be obtained based on this prediction probability.

[0081] In an alternative embodiment, for the case where the user does not know the specific vehicle information, the user can input a first vehicle image and a first question corresponding to the first vehicle image. At this time, the first question can be used to query whether the vehicle (or vehicle information) in the first vehicle image supports diagnosis. Then, in response to the user's input operation, the interaction device obtains the first vehicle image and the first question input by the user and sends the first question and the first vehicle image to the server.

[0082] At this time, the vehicle data supported by the diagnostic product for diagnosis includes text data and image data. The text data includes one or more of the brand, model, year, system, function, etc. of the vehicle supported by the diagnosis. The image data includes the vehicle image supported by the diagnosis. Then, based on the text data and the image data, the server generates a plurality of image-text pairs. Among them, there is an association relationship between the text and the image in each image-text pair. For example, there is a corresponding relationship between the vehicle information (such as the brand) and the text (such as the model) corresponding to the vehicle in the image. The details are not elaborated here. Then the server performs the following steps through the first model:

[0083] Extract features from the images in each image-text pair to obtain first image features corresponding to the images in each image-text pair, and extract features from the text in each image-text pair to obtain fifth features corresponding to the text in each image-text pair. Then perform attention processing on the first image features and the fifth features corresponding to each image-text pair to obtain sixth features corresponding to each image-text pair. Of course, before performing the attention processing, the first image features and the fifth features can also be mapped to the same dimensional space first.

[0084] Then extract features from the first vehicle image to obtain third image features; then generate a first answer based on the sixth features corresponding to each image-text pair, the third image features, and the tenth features (the principle is not elaborated here). For example, the sixth features corresponding to each image-text pair, the third image features, and the fourth features can be fused (such as concatenation, weighting) to obtain a first fused feature. Alternatively, the server can first determine the first similarity between the third image features and the sixth features corresponding to each image-text pair; then the server filters the plurality of image-text pairs based on the first similarity corresponding to each image-text pair to obtain a first image-text pair; then fuse the sixth features corresponding to the first image-text pair and the tenth features to obtain a first fused feature.

[0085] Then the server predicts the first fused feature through the first model to generate a first answer. For example, the first fused feature is processed according to Figure 1 the steps performed by the first model on the first feature vector in the embodiment to process the first fused feature to obtain a first answer. The specific principle is not elaborated here.

[0086] On the contrary, if the first matching result is not empty, the server performs entity recognition on the first question to obtain a third entity; then the server determines a fourth entity associated with the third entity based on the third entity and the vehicle data supported by the above vehicle diagnostic product. For example, the above Figure 3There is an association relationship between entities that are directly or indirectly connected in the embodiments. Then the server obtains the historical diagnostic data of the vehicle diagnostic product for the third entity and the fourth entity. For example, if the third entity and the fourth entity correspond to "Brand 1_Model 1", then the historical diagnostic data is the diagnostic data for the vehicle of "Brand 1_Model 1" (such as the total number of diagnoses, the systems and quantities diagnosed, the functions and quantities diagnosed, the quality feedback results of the diagnoses, etc.).

[0087] Then the server extracts features from the historical diagnostic data through the first model to obtain the seventh feature, and extracts features from the first matching result to obtain the fifteenth feature. Then the server generates the first answer based on the seventh feature and the fifteenth feature through the first model. For example, the seventh feature and the fifteenth feature are fused (not elaborated here), and the second fused feature is obtained; then the first answer is generated by predicting based on the second fused feature through the first model. For example, the second fused feature is processed according to Figure 1 The steps performed by the first model on the first feature vector in the embodiments are used to process the second fused feature to obtain the first answer, and the specific principle is not elaborated here. The first answer obtained at this time is the answer after adjusting and optimizing the first matching result. For example, the content of the first answer includes not only the answer to the first question but also diagnostic reference data. At this time, the diagnostic reference data is generated based on the historical diagnostic data, which can not only enrich the answer content but also provide diagnostic reference for users and increase the willingness of users to choose this diagnostic product.

[0088] The following introduces the flowchart of another interaction method of the present application. This method is applied to the server in the above embodiments; this method includes but is not limited to steps S21 - S22:

[0089] S21. The server obtains a first question, and the first question is used to query vehicle information supported by diagnosis.

[0090] In the embodiments of the present application, the user can input the first question through the interaction device. Then the interaction device obtains the first question in response to the user's input operation, and then the interaction device sends the first question to the server. Correspondingly, the server receives the first question from the interaction device.

[0091] S22. The server generates a first answer corresponding to the first question based on the first question and the first model, and the first model is trained with the vehicle data corresponding to the vehicles supported by the vehicle diagnostic product.

[0092] It should be noted that the specific principle of steps S21 - S22 can be referred to the corresponding explanations in steps S401 - S402 in the above embodiments, and will not be elaborated here.

[0093] Refer toFigure 6 , Figure 6 is a schematic diagram of the interaction process of an interaction method provided by an embodiment of the present application. This method is applied to the interaction system in the above embodiment; this method includes but is not limited to steps S601 - S605:

[0094] S601. The interaction device responds to the user's input operation and obtains a first question,

[0095] wherein the first question is used to query vehicle information supported by diagnosis.

[0096] S602. The interaction device sends the first question to the server.

[0097] S603. The server generates a first answer corresponding to the first question based on the first question and the first model.

[0098] Wherein, the first model is trained by vehicle data corresponding to vehicles supported by vehicle diagnosis products.

[0099] S604. The server sends the first answer to the interaction device.

[0100] S605. The interaction device displays the first answer.

[0101] It should be noted that the specific principles of steps S601 - S605 can be correspondingly referred to the corresponding explanations in the above embodiments, and will not be elaborated here.

[0102] It should be noted that the first model of the present application is trained by vehicle data corresponding to vehicles supported by vehicle diagnosis products. The training principle of the first model will be elaborated below in combination with specific embodiments as follows:

[0103] First, in combination with Figure 1 the structure of the first model in the embodiment, a training method of the first model is introduced. Refer to Figure 7 , Figure 7 is a training method of a first model provided by an embodiment of the present application. This method is applied to the server; this method includes but is not limited to steps S701 - S705:

[0104] S701. Obtain a second question.

[0105] In the embodiment of the present application, the second question is used as a training sample, and the number of training samples is not limited. In the embodiment of the present application, mainly one is taken as an example for explanation.

[0106] S702. Generate a first database based on vehicle data corresponding to vehicles supported by vehicle diagnosis products.

[0107] The production principle of the first database will not be elaborated here again. For details, please refer to the corresponding description in the above embodiments.

[0108] S703. Generate a second answer corresponding to the second question based on the second question and the first database.

[0109] Exemplarily, first, based on the second question, retrieve from the first database to obtain second data corresponding to the second question. The principle is similar to that of obtaining the first data above and will not be elaborated here. Then, based on the second question and the second data, generate a second prompt. The principle is similar to that of generating the first prompt above and will not be elaborated here. Then, based on the second prompt, generate a second answer. For example, input the second prompt into the first model, that is, use the second prompt as the input of the first model, and output a second answer corresponding to the second question. At this time, when using the second prompt as the input of the first model, the specific steps executed can refer to the steps executed by the first model for the input Input in the above Figure 1 embodiments and will not be elaborated here.

[0110] In an alternative embodiment, in combination with Figure 1 the embodiment, refer to Figure 8 , Figure 8 which is a schematic structural diagram of another first model provided by the embodiments of the present application. As Figure 8 shown, the first model includes N layers. Taking one layer as an example for explanation, it mainly includes a Feature Extraction Layer, Graph Convolutional Networks, Cross-Attention Layer, Masked Multi-head Attention, Feed-forward Neural Network, Normalization Layer Norm, Linear Layer, and Activation Layer Softmax. Among them, the Feature Extraction Layer includes a first feature extractor (i.e., Figure 8 FeatureExtraction(1) in Figure 8 ) and a second feature extractor (i.e.,

[0111] Therefore, in step S703, when generating a second answer corresponding to the second question based on the second question and the first database, the server may also generate multiple third answers based on the first database and multiple preset reply templates. The multiple third answers correspond to the multiple preset reply templates one by one, which will not be elaborated here. Then, based on the second question, the server determines a matching result corresponding to the second question from the multiple third answers. The principle is similar to that of the above first matching result and will not be elaborated here. Then, based on the matching result, the first database, and the second question, the server generates a second answer. If the matching result is empty, specifically:

[0112] First, the server performs entity annotation on the second question according to the entities in the knowledge graph to obtain a third question. Then, the third question is input into the first feature extractor Feature Extraction(1) for feature extraction to obtain a first feature, and the knowledge graph (or the corresponding graph structure) is input into the second feature extractor Feature Extraction(2) for feature extraction to obtain a second feature corresponding to the knowledge graph, which will not be elaborated here.

[0113] Then, the second feature is input into the graph convolutional network Graph Convolutional Networks for graph convolutional processing to obtain a third feature. Then, the first feature and the third feature are input into the cross-attention layer Cross-Attention Layer for cross-attention processing to obtain a fourth feature. Then, a second answer is generated based on the fourth feature. The principle is similar to that of generating a second answer based on the tenth feature above and can also be correspondingly referred to the above Figure 1 In the embodiment, the first model performs corresponding steps on the first feature vector to obtain an output Output, which will not be elaborated here. Among them, when generating a second answer based on the fourth feature, the fourth feature, the first feature, and the third feature can also be subjected to residual connection Add and normalization Norm processing to obtain a sixteenth feature, and then a second answer is generated based on the sixteenth feature, that is, according to Figure 1 the corresponding steps performed by the first model on the first feature vector in the embodiment.

[0114] In an alternative embodiment, based on Figure 9 the embodiment, refer to Figure 9 , Figure 9 which is a schematic structural diagram of another first model provided by the embodiment of the present application.

[0115] Figure 9The feature extraction layer in the first model shown also includes a third feature extractor Feature Extraction(3), and the third feature extractor includes a text feature extractor Text Feature and an image feature extractor ImageFeature. For the explanations of the remaining models, refer to Figure 8 the explanations in the embodiments, which will not be elaborated here. At this time, the above vehicle data may include text data and image data. The text data includes one or more of the brand, model, year, system, function, etc. of the vehicle that supports diagnosis, and the image data includes vehicle images that support diagnosis. Correspondingly, then in terms of generating the second answer based on the fourth feature,

[0116] the server generates multiple image-text pairs based on the text data and image data. Among them, there is an association relationship between the text and the image in each image-text pair, which will not be elaborated here. Then, the image feature extractor in the third feature extractor of the first model extracts features from the images in each image-text pair to obtain first image features corresponding to the images in each image-text pair, and the text feature extractor in the third feature extractor of the first model extracts features from the text in each image-text pair to obtain fifth features corresponding to the text in each image-text pair.

[0117] Then, the first image features and fifth features corresponding to each image-text pair are input into a cross-attention layer Cross-Attention Layer for cross-attention processing to obtain sixth features corresponding to each image-text pair. Then, the server obtains the first vehicle image corresponding to the first question and inputs the first vehicle image into the image feature extractor in the third feature extractor for feature extraction to obtain second image features.

[0118] Then, based on the sixth features, second image features, and fourth features corresponding to each image-text pair, a second answer is generated. For example, Figure 9 as shown, the sixth features, second image features, and fourth features corresponding to each image-text pair are fused to obtain a third fused feature; alternatively, optionally, multiple image-text pairs may be filtered based on the second image features and the sixth features corresponding to each image-text pair to obtain a first image-text pair. The principle will not be elaborated here. Then, the sixth features and fourth features corresponding to the first image-text pair are fused to obtain a third fused feature. This application does not make a limitation.

[0119] Then, the third fused feature, the first image features, fifth features, first features, and third features corresponding to each image-text pair are subjected to residual connection Add and then normalization Norm processing to obtain seventeenth features. Then, a second answer is generated based on the seventeenth features. The principle is similar to the principle of generating the second answer based on the sixteenth features, orFigure 1 The corresponding steps performed by the first model on the first feature vector in the embodiments are processed and will not be elaborated here.

[0120] Conversely, if the matching result is not empty, when the server generates a second answer based on the matching result, the first database, and the second question, it first performs entity recognition on the second question to obtain a first entity; then based on the first entity and vehicle data, it determines a second entity that has an association relationship with the first entity, which will not be elaborated here; then it obtains the historical diagnostic data of the vehicle diagnostic product for the first entity and the second entity, which will not be elaborated here; then it extracts features from the historical diagnostic data through the first feature extractor in the first model to obtain a seventh feature, and extracts features from the matching result through the first feature extractor in the first model to obtain an eighth feature; then based on the seventh feature and the eighth feature, it obtains a second answer. For example, it inputs the seventh feature and the eighth feature into a cross-attention layer for attention processing to obtain an eighteenth feature; then it performs residual connection Add and normalization Norm processing on the eighteenth feature, the seventh feature, and the eighth feature to obtain a nineteenth feature; then it generates a second answer based on the nineteenth feature. The principle is similar to the principle of generating the second answer based on the sixteenth feature above, or in accordance with Figure 1 The corresponding steps performed by the first model on the first feature vector in the embodiments are processed and will not be elaborated here.

[0121] S704. Determine the training loss based on the second answer.

[0122] After generating the second answer, the training loss can be calculated based on the second answer and the true label corresponding to the training sample, that is, the second question, such as cross-entropy loss, mean squared error loss, etc. This application does not limit the type of loss, and the training loss is obtained.

[0123] S705. Train the first model based on the training loss.

[0124] Then, based on the training loss, the model can be trained. For example, all parameters of the first model can be fine-tuned, or some parameters of the first model can be fine-tuned, such as the weights in masked multi-head attention and cross-attention. For example, the LoRA fine-tuning method can be used. This application does not limit it until the first model converges, and the trained first model is obtained.

[0125] Furthermore, after training the first model until it converges based on the above embodiments, the trained first model can be applied. The following introduces the main application scenarios involved in this application in conjunction with the accompanying drawings. This scenario includes a server and an interaction device. The first model is deployed on the server, specifically as follows:

[0126] Combined with Figure 5Example, refer to Figure 10 , Figure 10 which is a schematic diagram of a scenario provided by an embodiment of the present application. As Figure 10 shown, the user enters the question "Is the 2020 BMW X5 supported for diagnosis?" in the input box displayed on the first interface of the target application or the target web page on the interaction device. At this time, this question is used to query vehicle information supported for diagnosis, that is, the brand is BMW, the model is X5, and the year is 2020. And the user selects the first function button corresponding to function 2 (at this time, this function button is in a selected state, as Figure 10 shown by the shaded filling), and the intention corresponding to the first function button is "used to determine vehicle information supported for diagnosis". Then the user can click the "Send" function button, and the interaction device responds to this operation and sends the question and intention to the server.

[0127] Correspondingly, the interaction device can display the question entered by the user ( Figure 10 not shown), and that the server has received the question and intention; then the server, through the first model, based on the question and the database corresponding to this intention, generates an answer corresponding to this question. The specific principle will not be elaborated here. Figure 10 The example of the answer shown is "For your BMW X5 model, the model years that our device can diagnose range from 1998 to 2024. Therefore, the 2020 BMW X5 is supported for diagnosis. In addition to the X5, we also support diagnosing other BMW models, such as the 1 Series, 2 Series, 3 Series, etc., a total of about 29 models. If you have questions about other models or model years, welcome to continue asking."; then the server sends this answer to the interaction device. Correspondingly, the interaction device receives this answer and displays this answer, presenting Figure 10 a round of conversation as shown.

[0128] Of course, optionally, refer to Figure 11 , Figure 11 which is another schematic diagram of a scenario provided by an embodiment of the present application. As Figure 11 shown, the user can also not select the function button and only enter the question "Is the 2020 BMW X5 supported for diagnosis?" in the input box. At this time, this question is used to query vehicle information supported for diagnosis, that is, the brand is BMW, the model is X5, and the year is 2020. Then the user can click the "Send" function button, and the interaction device responds to this operation and sends this question to the server.

[0129] Correspondingly, the interaction device can display the question entered by the user ( Figure 11 not shown), and that the server has received this question; then the server determines the corresponding intention based on this question; then the server, through the first model, based on the question and the database corresponding to this intention, generates an answer corresponding to this question. The specific principle will not be elaborated here.Figure 11 The example answer shown is "According to professional diagnostic knowledge, the model years supported by the BMW X5 are from 1998 to 2024. Therefore, the BMW X5 in 2020 is supported for diagnosis. In addition to the BMW X5, there are many other models and supported model years in the BMW series, such as the 1 Series (2003 - 2024), 3 Series (1996 - 2024), 5 Series (1996 - 2024), etc. There are approximately 29 models."; then the server sends this answer to the interaction device. Correspondingly, the interaction device receives this answer and displays it, presenting Figure 11 the round of conversation shown.

[0130] It should be noted that Figure 10 and Figure 11 The content shown in the embodiments, such as questions, answers, page layouts, etc., are only examples. Making corresponding extensions or variations based on this, and applying the interaction method of this application in other scenarios, all fall within the protection scope of this application.

[0131] Referring to Figure 12 , Figure 12 is a block diagram of the functional units of an interaction device provided in an embodiment of this application. The interaction device 1200 includes: a first acquisition unit 1201 and a first processing unit 1202;

[0132] The first acquisition unit 1201 is configured to obtain a first question in response to a user's input operation, where the first question is used to query vehicle information supported for diagnosis;

[0133] The first processing unit 1202 is configured to display a first answer corresponding to the first question, where the first answer is obtained based on the first question and a first model, and the first model is trained using vehicle data corresponding to vehicles supported by a vehicle diagnostic product.

[0134] In specific implementation, the first acquisition unit 1201 and the first processing unit 1202 described in the embodiments of the present invention may also perform other manners executed by the interaction device in other interaction method embodiments provided by the embodiments of the present invention, which will not be elaborated here.

[0135] Referring to Figure 13 , Figure 13 is a block diagram of the functional units of a server provided in an embodiment of this application. The server 1300 includes: a second acquisition unit 1301 and a second processing unit 1302;

[0136] The second acquisition unit 1301 is configured to obtain a first question, where the first question is used to query vehicle information supported for diagnosis;

[0137] A second processing unit 1302, configured to generate a first answer corresponding to the first question based on the first question and a first model, where the first model is trained using vehicle data corresponding to a vehicle supported by a vehicle diagnostic product.

[0138] In a specific implementation, the second obtaining unit 1301 and the second processing unit 1302 described in the embodiments of the present invention may also execute other operations performed by the server in the embodiments of the interaction method provided by the embodiments of the present invention, which will not be elaborated herein.

[0139] Refer to Figure 14 , Figure 14 which is a schematic structural diagram of an electronic device provided by an embodiment of the present application. As Figure 14 shown, the electronic device 1400 includes a transceiver 1401, a processor 1402, and a memory 1403. They are connected through a bus 1404. The memory 1403 is used to store computer programs and data, and can transmit the data stored in the memory 1403 to the processor 1402.

[0140] The electronic device 1400 may be the interaction device 1200 or the server 1300;

[0141] When the electronic device 1400 is the interaction device 1200, the processor 1402 is configured to read the computer program in the memory 1403 and perform the following operations:

[0142] Control the transceiver 1401 to obtain a first question in response to a user's input operation, where the first question is used to query vehicle information supported by diagnosis;

[0143] Display a first answer corresponding to the first question, where the first answer is obtained based on the first question and a first model, and the first model is trained using vehicle data corresponding to a vehicle supported by a vehicle diagnostic product.

[0144] In a specific implementation, the transceiver 1401 and the processor 1402 described in the embodiments of the present invention may also execute other implementation manners described in the embodiments of the interaction method provided by the embodiments of the present invention, which will not be elaborated herein.

[0145] When the electronic device 1400 is the server 1300, the processor 1402 is configured to read the computer program in the memory 1403 and perform the following operations:

[0146] Control the transceiver 1401 to obtain a first question, where the first question is used to query vehicle information supported by diagnosis;

[0147] Generate a first answer corresponding to the first question based on the first question and a first model, where the first model is trained using vehicle data corresponding to a vehicle supported by a vehicle diagnostic product.

[0148] In a specific implementation, the transceiver 1401 and the processor 1402 described in the embodiments of the present invention may also implement other implementation manners described in the embodiments of the interaction method provided by the embodiments of the present invention, which will not be elaborated herein.

[0149] Specifically, the above transceiver 1401 may be Figure 12 the first acquisition unit 1201 of the interaction device 1200 in the embodiment of Figure 13 or the second acquisition unit 1301 of the server 1300 in the embodiment of Figure 12 and the above processor 1402 may be Figure 13 the first processing unit 1202 of the interaction device 1200 in the embodiment of

[0150] It should be understood that the embodiments of the present application also provide a computer-readable storage medium, which stores a computer program. The computer program is executed by a processor to implement part or all of the steps of any one of the interaction methods described in the above method embodiments.

[0151] The embodiments of the present application also provide a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program. The computer program is operable to cause a computer to execute part or all of the steps of any one of the interaction methods described in the above method embodiments.

[0152] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present application is not limited by the described action sequence, because according to the present application, certain steps may be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present application.

[0153] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not elaborated in a certain embodiment, reference may be made to the relevant descriptions of other embodiments.

[0154] In several embodiments provided by this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be in electrical or other forms.

[0155] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0156] In addition, each functional unit in various embodiments of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software program modules.

[0157] If the integrated unit is implemented in the form of a software program module and sold or used as an independent product, it can be stored in a computer-readable memory. Based on such an understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods in various embodiments of this application. And the aforementioned memory includes: USB flash drives, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), mobile hard disks, magnetic disks, or optical discs and other media that can store program codes.

[0158] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program. This program can be stored in a computer-readable memory. The memory can include: flash drives, read-only memories (abbreviation: ROM), random access memories (abbreviation: RAM), magnetic disks, or optical discs, etc.

[0159] The embodiments of the present application have been introduced in detail above. Specific examples are used in this text to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to the present application.

Claims

1. An interactive method, characterized in that: The method comprises: In response to an input operation of a user, obtaining a first question, wherein the first question is used to query vehicle information supporting diagnosis; A first answer corresponding to the first question is displayed, wherein the first answer is obtained based on the first question and a first model, and the first model is trained with vehicle data corresponding to vehicles supported for diagnosis by a vehicle diagnostic product.

2. The method according to claim 1, characterized in that Before displaying the first answer corresponding to the first question, the method further includes executing the following steps by the server: Acquire a first intention of the user, where the first intention is used to determine vehicle information supporting diagnosis; The first answer is obtained based on the first question, a first database corresponding to the first intention, and the first model, wherein the first database is generated based on vehicle data corresponding to a vehicle supported for diagnosis by a vehicle diagnostic product.

3. The method according to claim 2, characterized in that The obtaining the first intention of the user includes: In response to a touch operation of the user on a first function button, acquiring a first intention of the user; or, Perform keyword recognition on the first question to obtain a first keyword; Based on the first keyword, the first intent is determined.

4. The method according to any one of claims 1 to 3, characterized in that: The training method of the first model comprises the following steps: Get the second question; Based on the vehicle data, generating a first database; Based on the second question and the first database, generating a second answer corresponding to the second question; Based on the second answer, determining a training loss; The first model is trained based on the training loss.

5. The method according to claim 4, characterized in that The generating a second answer corresponding to the second question based on the second question and the first database includes: Based on the first database and a plurality of preset answer templates, generating a plurality of third answers, wherein the plurality of third answers correspond one-to-one to the plurality of preset answer templates; Based on the second question, determining a matching result corresponding to the second question from the plurality of third answers; The second answer is generated based on the matching result, the first database and the second question.

6. The method according to claim 5, characterized in that The first database is a knowledge graph generated based on the vehicle data; If the matching result is empty, generating the second answer based on the matching result, the first database and the second question includes: According to the entities in the knowledge graph, entity labeling is performed on the second question to obtain a third question; Performing feature extraction on the third question to obtain a first feature, and performing feature extraction on the knowledge graph to obtain a second feature corresponding to the knowledge graph; Performing graph convolution processing on the second feature to obtain a third feature; Performing attention processing based on the first feature and the third feature to obtain a fourth feature; Based on the fourth feature, the second answer is generated.

7. The method according to claim 5, characterized in that If the matching result is not empty, generating the second answer based on the matching result, the first database and the second question includes: Performing entity recognition on the second question to obtain a first entity; Based on the first entity and the vehicle data, determining a second entity associated with the first entity; Acquire historical diagnostic data between the first entity and the second entity by the vehicle diagnostic product; Performing feature extraction on the historical diagnostic data to obtain a seventh feature, and performing feature extraction on the matching result to obtain an eighth feature; Based on the seventh feature and the eighth feature, the second answer is obtained.

8. An interactive device, characterized in that: The interaction device comprises: a first acquisition unit and a first processing unit; The first acquisition unit is used to acquire a first question in response to an input operation of a user, wherein the first question is used to query vehicle information supporting diagnosis; The first processing unit is used to display a first answer corresponding to the first question, wherein the first answer is obtained based on the first question and a first model, and the first model is trained with vehicle data corresponding to a vehicle supported by a vehicle diagnostic product.

9. An electronic device, characterized in that: include: A processor and a memory, wherein the processor is connected to the memory, the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the method according to any one of claims 1 to 7.

Images