A data processing method, device, computer, storage medium and program product

Abstract

The embodiment of the present application discloses a data processing method, device, computer, storage medium and program product, which relates to data transmission technology in the field of big data. The method includes: inputting business processing instructions and virtual asset-related data of aircraft into a target business processing model; based on asset data classification rules and components, performing data division on virtual asset-related data to obtain S units of virtual assets, determining the binary classification information corresponding to each unit of virtual assets and obtaining the corresponding weight model parameters; combining data feature vectors and weight model parameters to obtain S fused feature vectors; generating prompt text based on business processing instructions and determining a target processing network, performing feature processing on S fused feature vectors and prompt text through the target processing network to obtain feature processing results; classifying and identifying the feature processing results to obtain data recognition results. The present application can improve the accuracy of data recognition results.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a data processing method, device, computer, storage medium and program product. Background Art

[0002] Classification problems are widely used in various fields, such as image classification, natural language processing, medical diagnosis, etc. Classification problems can also be applied to data type identification and data processing. Traditional data classification or analysis methods require manual support. In the case of a wide variety of data sources, manual analysis is not only inefficient and costly, but also difficult to find important data in different types of data sources. In addition, when the data categories are complex, the accuracy of the final data classification or analysis results is low. Summary of the invention

[0003] The embodiments of the present application provide a data processing method, device, computer, storage medium and program product, which can achieve targeted processing of data through asset data classification rules and business processing instructions, improve business processing efficiency, save time, and improve the accuracy of data identification results.

[0004] On the one hand, an embodiment of the present application provides a data processing method, the method comprising:

[0005] Acquire the business processing instruction and the virtual asset associated data for the aircraft sent by the first business object, and input the business processing instruction and the virtual asset associated data for the aircraft into the target business processing model; the aircraft is composed of at least two components;

[0006] Obtain asset data classification rules, and divide the virtual asset associated data of the aircraft based on N business types and at least two components in the asset data classification rules to obtain S unit virtual assets, and determine the binary classification information corresponding to each unit virtual asset; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers;

[0007] Based on the binary classification information corresponding to the S unit virtual assets, respectively, the weight model parameters corresponding to the S unit virtual assets are obtained from the weight model parameter set in the target business processing model; the weight model parameter set includes H weight model parameters that respectively represent different impact weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer;

[0008] The data feature vectors corresponding to the S units of virtual assets and the weight model parameters corresponding to the S units of virtual assets are combined to obtain fused feature vectors corresponding to the S units of virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter;

[0009] Generate a prompt text for indicating the identification requirement type based on the business processing instruction, determine a target processing network corresponding to the identification requirement type from a target business processing model based on the prompt text, perform feature processing on the S fused feature vectors and the prompt text through the target processing network, and obtain a feature processing result;

[0010] The feature processing results are classified and identified to obtain data identification results for responding to business processing instructions.

[0011] On the one hand, an embodiment of the present application provides another data processing method, the method comprising:

[0012] Obtaining a sample processing instruction, sample data of virtual assets for a sample aircraft, and a sample identification result, and inputting the sample processing instruction and the sample data of virtual assets for the sample aircraft into an initial business processing model; the sample aircraft is composed of at least two sample components;

[0013] Obtain asset data classification rules, and divide the virtual asset sample data of the sample aircraft based on N business types and at least two sample components in the asset data classification rules to obtain R unit sample virtual assets, and determine the sample binary classification information corresponding to each unit sample virtual asset; the sample binary classification information is used to characterize the business type and sample component to which the unit sample virtual asset belongs; N and R are both positive integers;

[0014] Assign random initialization model parameters to the binary classification information corresponding to the R unit sample virtual assets respectively, and determine the R initialization model parameters as the initialization model parameter set;

[0015] The sample feature vectors corresponding to the R unit sample virtual assets and the initialization model parameters corresponding to the R unit sample virtual assets are combined to obtain the sample fusion feature vectors corresponding to the R unit sample virtual assets; a sample fusion feature vector is composed of a sample feature vector and an initialization model parameter;

[0016] Generate a sample prompt text for indicating the sample identification requirement type based on the sample processing instruction, determine a sample processing network corresponding to the sample identification requirement type from the initial business processing model based on the sample prompt text, perform feature processing on R sample fusion feature vectors and the sample prompt text through the sample processing network to obtain a sample feature processing result, classify and identify the sample feature processing result, and obtain a sample prediction result for responding to the sample processing instruction;

[0017] Based on the sample prediction results and the sample recognition results, a model loss value is generated, and each initialization model parameter in the initialization model parameter set and the parameters of the sample processing network are adjusted based on the model loss value until the initial business processing model meets the model convergence conditions, thereby obtaining a target business processing model; the target business processing model includes a weight model parameter set, and the weight model parameter set includes weight model parameters obtained after adjusting the initialization model parameters.

[0018] An embodiment of the present application provides a data processing device, the device comprising:

[0019] A data acquisition module, used to acquire the business processing instruction sent by the first business object and the virtual asset associated data for the aircraft, and input the business processing instruction and the virtual asset associated data of the aircraft into the target business processing model; the aircraft is composed of at least two components;

[0020] The data partitioning module is used to obtain the asset data classification rules, and based on the N business types and at least two components in the asset data classification rules, the virtual asset associated data of the aircraft is partitioned to obtain S unit virtual assets, and the binary classification information corresponding to each unit virtual asset is determined; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers;

[0021] A parameter acquisition module is used to acquire the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model based on the binary classification information corresponding to the S unit virtual assets; the weight model parameter set includes H weight model parameters representing different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer;

[0022] A feature combination module is used to combine the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets to obtain fused feature vectors corresponding to the S unit virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter;

[0023] A network determination module is used to generate a prompt text indicating the identification requirement type based on the business processing instruction, determine a target processing network corresponding to the identification requirement type from a target business processing model based on the prompt text, and perform feature processing on the S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result;

[0024] The classification and recognition module is used to classify and recognize the feature processing results to obtain data recognition results for responding to business processing instructions.

[0025] In a possible implementation, the data partitioning module is used to partition the virtual asset associated data of the aircraft based on the N business types and at least two components in the asset data classification rule to obtain S unit virtual assets. When determining the binary classification information corresponding to each unit virtual asset, the data partitioning module is specifically used to perform the following operations:

[0026] The virtual asset associated data of the aircraft is classified according to the N business types in the asset data classification rule to obtain M types of classified data; the business types corresponding to the M types of classified data belong to the N business types, and M is a positive integer less than or equal to N;

[0027] In each type of classified data, each type of classified data is divided into components based on at least two components to obtain S unit virtual assets; one unit virtual asset corresponds to one business type and one component;

[0028] The business types and components corresponding to the S unit virtual assets are obtained respectively, and the business types and components corresponding to the S unit virtual assets are combined to obtain the binary classification information corresponding to the S unit virtual assets respectively.

[0029] In a possible implementation, the N types of business types include asset business data types, basic business data types, and transaction business data types; if M is equal to N, the M types of classified data include first classified data, second classified data, and third classified data; the data partitioning module is used to classify the virtual asset associated data of the aircraft according to the N types of business types in the asset data classification rule, and when the M types of classified data are obtained, the data partitioning module is specifically used to perform the following operations:

[0030] Obtain the data key fields corresponding to the asset business data type; the data key fields include the usage record key fields, maintenance record key fields, and impairment record key fields;

[0031] In the virtual asset associated data of the aircraft, historical use records corresponding to the use record key field, historical maintenance records corresponding to the maintenance record key field, and historical damage records corresponding to the damage record key field are obtained, and the historical use records, historical maintenance records, and historical damage records are determined as first classification data corresponding to the asset business data type;

[0032] Obtain the factory key field corresponding to the basic business data type, obtain the factory associated data corresponding to the factory key field in the virtual asset associated data of the aircraft, and determine the factory associated data as the second classification data corresponding to the basic business data type;

[0033] Obtain the transaction key field corresponding to the transaction business data type, obtain the transaction related data corresponding to the transaction key field in the virtual asset related data of the aircraft, and determine the transaction related data as the third classification data corresponding to the transaction business data type.

[0034] In a possible implementation, the binary classification information corresponding to the S units of virtual assets includes binary classification information B i , i is a positive integer less than or equal to S; when the parameter acquisition module is used to obtain the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model based on the binary classification information corresponding to the S unit virtual assets, the parameter acquisition module is specifically used to perform the following operations:

[0035] Obtaining a weight model parameter set and a parameter matching rule in a target business processing model; the parameter matching rule is used to indicate a mapping relationship between a weight model parameter in the weight model parameter set and binary classification information;

[0036] Determine a binary classification information set based on a weight model parameter set and a parameter matching rule;

[0037] Combine the binary classification information set with the binary classification information B i The matched bigram classification information is determined as the target bigram classification information;

[0038] Based on the target binary classification information and parameter matching rules, the target weight model parameters corresponding to the target binary classification information are obtained from the weight model parameters, and the target weight model parameters are determined as the binary classification information B i The corresponding weight model parameters.

[0039] In a possible implementation, the S unit virtual assets include unit virtual asset A i , the data feature vectors corresponding to the S unit virtual assets include unit virtual asset A i The corresponding data feature vector Ci , the weight model parameters corresponding to the S units of virtual assets include unit virtual asset A i The corresponding weight model parameter D i , i is a positive integer less than or equal to S; the feature combination module is used to combine the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets, and when the fused feature vectors corresponding to the S unit virtual assets are obtained, the feature combination module is specifically used to perform the following operations:

[0040] Determine the identification requirement type and unit virtual asset A indicated by the business processing instruction i The correlation between virtual assets A based on the correlation i Assigning data association weights;

[0041] For the data feature vector C i and weight model parameter D i Splice and obtain data splicing features, and multiply the data splicing features and the data association weight to determine the unit virtual asset A i The corresponding fused feature vector.

[0042] In a possible implementation, the target processing network is a pricing prediction network; the network determination module is used to perform feature processing on the S fused feature vectors and the prompt text through the target processing network, and when the feature processing result is obtained, the network determination module is specifically used to perform the following operations:

[0043] Extract features of the prompt text to obtain prompt text features;

[0044] Based on the search engine associated with the pricing prediction network, the environmental factor information of the object to be priced indicated by the prompt text is obtained;

[0045] Extract features of environmental factor information to obtain environmental factor features, fuse the environmental factor features, prompt text features and S fused feature vectors to obtain feature processing results;

[0046] The classification and recognition module is used to classify and recognize the feature processing results. When the data recognition results for responding to the business processing instructions are obtained, the classification and recognition module is specifically used to perform the following operations:

[0047] Obtain P price range labels corresponding to the pricing prediction network, and the estimated market information corresponding to the P price range labels, respectively, perform feature extraction on the P estimated market information, and obtain P price market features; P is a positive integer;

[0048] Perform cross-attention processing on the feature processing results and P price market features to obtain P first cross-attention scores;

[0049] The price range label corresponding to the estimated market information associated with the first target cross-attention score is determined as the data recognition result for responding to the business processing instruction; the first target cross-attention score is the largest first cross-attention score among P first cross-attention scores.

[0050] In a possible implementation, the classification and recognition module is used to perform cross-attention processing on the feature processing result and P price market features. When P first cross-attention scores are obtained, the classification and recognition module is specifically used to perform the following operations:

[0051] Combine P price market features to obtain price combination features;

[0052] The feature processing result is used as the query vector in the cross attention function, the price combination feature is used as the key vector in the cross attention function, and the price combination feature is used as the value vector in the cross attention function;

[0053] The product of the query vector and the key vector is determined as the first fused feature through the cross attention function;

[0054] The number of dimensions corresponding to the price combination feature is obtained, and the product of the first fusion feature and the inverse of the number of dimensions is determined as the second fusion feature;

[0055] Based on the activation sub-function in the cross attention function, the second fused feature is converted into the first activated feature;

[0056] The product of the first activation feature and the value vector determines the attention processing sequence; the attention processing sequence is composed of P first cross-attention scores.

[0057] In a possible implementation, the target processing network is a data analysis network; the network determination module is used to perform feature processing on the S fused feature vectors and the prompt text through the target processing network, and when the feature processing result is obtained, the network determination module is specifically used to perform the following operations:

[0058] Extract features of the prompt text to obtain prompt text features;

[0059] Fuse the prompt text features and S fused feature vectors to obtain feature processing results;

[0060] The classification and recognition module is used to classify and recognize the feature processing results. When the data recognition results for responding to the business processing instructions are obtained, the classification and recognition module is specifically used to perform the following operations:

[0061] Obtain Q data classification labels corresponding to the data analysis network, and label association information corresponding to the Q data classification labels, respectively, perform feature extraction on the Q label association information, and obtain Q data classification features; Q is a positive integer;

[0062] Perform cross-attention processing on the feature processing results and Q data classification features to obtain Q second cross-attention scores;

[0063] The data classification label corresponding to the label association information associated with the second target cross-attention score is determined as the data recognition result for responding to the business processing instruction; the second target cross-attention score is the largest second cross-attention score among the Q second cross-attention scores.

[0064] In a possible implementation, the data processing device further includes a result determination module, which is specifically configured to perform the following operations:

[0065] Perform information retrieval on the data recognition results and prompt texts through a large language model to obtain information retrieval results;

[0066] Acquire a template library associated with the target processing network, and acquire a service template corresponding to the prompt text from the template library;

[0067] Determine the text to be updated in the business template, replace the text to be updated in the business template based on the data recognition result and the information retrieval result to obtain the updated business text, and determine the updated business text as the business result for the business processing instruction.

[0068] In a possible implementation, the result determination module is further configured to perform the following operations:

[0069] If the identification requirement type indicated by the prompt text includes a drawing processing requirement, the drawing image type is identified according to the drawing processing requirement, an image template having the drawing image type is obtained from an image template library, and a Gaussian noise image is generated according to the image template and the initial noise data;

[0070] The data recognition result and the prompt text are input into the large language model, and the data recognition result and the prompt text are fused and expanded by the large language model to obtain the drawing instruction text;

[0071] The Gaussian noise image and the drawing instruction text are input into the Vincent graph model, the Gaussian noise image is feature extracted by the Vincent graph model to obtain Gaussian noise features, and the Gaussian noise features are forward diffused to obtain a forward noise vector;

[0072] The drawing instruction text is feature-encoded by the text-generated graph model to obtain the drawing text feature, and the Gaussian noise image is denoised according to the forward noise vector and the drawing text feature to obtain the drawing data graph;

[0073] The data identification result and the drawn data graph are determined as business results for the business processing instruction.

[0074] On the one hand, an embodiment of the present application provides another data processing device, which includes:

[0075] A sample acquisition module, used to acquire sample processing instructions, virtual asset sample data for a sample aircraft, and sample identification results, and input the sample processing instructions and the virtual asset sample data of the sample aircraft into an initial business processing model; the sample aircraft is composed of at least two sample components;

[0076] The sample partitioning module is used to obtain the asset data classification rules, and based on the N business types and at least two sample components in the asset data classification rules, the virtual asset sample data of the sample aircraft is partitioned to obtain R unit sample virtual assets, and the sample binary classification information corresponding to each unit sample virtual asset is determined; the sample binary classification information is used to characterize the business type and sample component to which the unit sample virtual asset belongs; N and R are both positive integers;

[0077] A parameter allocation module, used to allocate random initialization model parameters to the binary classification information corresponding to the R unit sample virtual assets, and determine the R initialization model parameters as an initialization model parameter set;

[0078] The sample feature combination module is used to combine the sample feature vectors corresponding to the R unit sample virtual assets and the initialization model parameters corresponding to the R unit sample virtual assets to obtain the sample fusion feature vectors corresponding to the R unit sample virtual assets; a sample fusion feature vector is composed of a sample feature vector and an initialization model parameter;

[0079] A sample feature processing module is used to generate a sample prompt text indicating the sample identification requirement type based on the sample processing instruction, determine a sample processing network corresponding to the sample identification requirement type from the initial business processing model based on the sample prompt text, perform feature processing on R sample fusion feature vectors and the sample prompt text through the sample processing network to obtain a sample feature processing result, classify and identify the sample feature processing result, and obtain a sample prediction result for responding to the sample processing instruction;

[0080] The parameter adjustment module is used to generate a model loss value based on the sample prediction results and the sample recognition results, and to adjust each initialization model parameter in the initialization model parameter set and the parameters of the sample processing network based on the model loss value until the initial business processing model satisfies the model convergence conditions and obtains the target business processing model; the target business processing model includes a weight model parameter set, and the weight model parameter set includes the weight model parameters obtained after the initialization model parameters are adjusted.

[0081] On the one hand, an embodiment of the present application provides a computer device, including a processor, a memory, and an input and output interface;

[0082] The processor is connected to the memory and the input / output interface respectively, wherein the input / output interface is used to receive and output data, the memory is used to store a computer program, and the processor is used to call the computer program so that a computer device including the processor executes the method in one aspect of an embodiment of the present application.

[0083] On the one hand, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program, and the computer program is suitable for being loaded and executed by a processor so that a computer device having the processor executes the method in one aspect of the embodiment of the present application.

[0084] In one aspect, an embodiment of the present application provides a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the method provided in various optional ways in one aspect of the embodiment of the present application. In other words, when the computer instruction is executed by the processor, the method provided in various optional ways in one aspect of the embodiment of the present application is implemented.

[0085] Implementing the embodiments of the present application will have the following beneficial effects:

[0086] In an embodiment of the present application, a business processing instruction sent by a first business object and virtual asset associated data for an aircraft are obtained, and the business processing instruction and the virtual asset associated data of the aircraft are input into a target business processing model; the aircraft is composed of at least two components; an asset data classification rule is obtained, and based on N business types and at least two components in the asset data classification rule, the virtual asset associated data of the aircraft is divided into data to obtain S unit virtual assets, and the binary classification information corresponding to each unit virtual asset is determined; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers; based on the binary classification information corresponding to the S unit virtual assets, the weight model parameters corresponding to the S unit virtual assets are obtained from the weight model parameter set in the target business processing model; the weight The model parameter set includes H weight model parameters respectively representing different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer; the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets are combined to obtain fused feature vectors corresponding to the S unit virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter; a prompt text for indicating the identification requirement type is generated based on the business processing instruction, a target processing network corresponding to the identification requirement type is determined from the target business processing model based on the prompt text, and feature processing is performed on the S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result; the feature processing result is classified and identified to obtain a data identification result for responding to the business processing instruction. Through the above process, the asset data classification rules and business processing instructions are used to implement targeted processing of virtual asset-related data, which can improve business processing efficiency and save time; the virtual asset-related data is divided according to the asset data classification rules, and the business type and component (binary classification information) to which the unit virtual asset belongs can be obtained; the weight information (weight model parameters) of the unit virtual asset binary classification information is fully utilized, and the weight model parameters can be used to better find important data suitable for the identification requirement type, thereby increasing the influence ratio of important data, being able to better cope with business processing problems in complex data situations, and improving the recognition accuracy of the final data recognition results. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0088] Figure 1 This is a network interaction architecture diagram provided by an embodiment of the present application;

[0089] Figure 2 It is a scenario schematic diagram of a data processing method provided in an embodiment of the present application;

[0090] Figure 3 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 1 ;

[0091] Figure 4 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 2 ;

[0092] Figure 5 This is a data classification diagram provided in an embodiment of the present application;

[0093] Figure 6 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 3 ;

[0094] Figure 7 A data processing device provided in an embodiment of the present application is shown in FIG. Figure 1 ;

[0095] Figure 8 A data processing device provided in an embodiment of the present application is shown in FIG. Figure 2 ;

[0096] Fig. 9 It is a structural diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0097] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0098] Among them, if it is necessary to collect object (such as user, etc.) data in this application, a prompt interface or pop-up window will be displayed before or during the collection. The prompt interface or pop-up window is used to prompt the user that certain data is currently being collected. Only after the user confirms the prompt interface or pop-up window, the relevant steps of data acquisition will be started, otherwise the process will end. Moreover, the acquired user data will be used in reasonable and legal scenarios or purposes. Optionally, in some scenarios where user data needs to be used but has not been authorized by the user, authorization can be requested from the user, and the user data can be used when the authorization is passed.

[0099] It is understandable that in the specific implementation of this application, the user data involved, when the following embodiments of this application are applied to specific products or technologies, needs to obtain user permission or consent, and the collection, use and processing of relevant data need to comply with relevant laws, regulations and standards of the relevant regions.

[0100] In the examples of this application, see Figure 1 , Figure 1 is a network interaction architecture diagram provided by an embodiment of the present application, such as Figure 1 As shown, the network interaction architecture diagram may include a business server 101 and a terminal device cluster, and the terminal device cluster may include terminal device 102a, terminal device 102b, terminal device 102c, ..., terminal device 102n, wherein there may be a communication connection between the terminal device clusters, for example, there is a communication connection between terminal device 102a and terminal device 102b, and there is a communication connection between terminal device 102a and terminal device 102c. At the same time, any terminal device in the terminal device cluster may have a communication connection with the business server 101, for example, there is a communication connection between terminal device 102a and business server 101, wherein the above communication connection does not limit the connection method, and may be directly or indirectly connected through a wired communication method, or directly or indirectly connected through a wireless communication method, or through other methods, and this application does not make any restrictions here.

[0101] It should be understood that Figure 1 Each terminal device in the terminal device cluster shown in FIG. 1 may be installed with an application client having a data processing function. When the application client runs in each terminal device, an intelligent conversation page may be displayed and the above-mentioned Figure 1 The business servers 101 shown in the figure interact with each other so that the business server 101 can receive business data from each terminal device. Among them, the application client can be a social application, instant messaging application, live broadcast application, short video application, video application, music application, shopping application, novel application, browser, etc., which has the function of displaying text, images, audio, video and other data information. Among them, the application client can be an independent client or an embedded sub-client (for example, application applet, browser web page access, etc.) integrated in a client (for example, instant messaging client, social client, video client, etc.), which is not limited here.

[0102] like Figure 1As shown, any terminal device in the terminal device cluster can send business processing instructions and virtual asset associated data for the aircraft to the business server 101. Here, taking the terminal device 102a as the terminal device corresponding to the first business object as an example, the business server 101 can obtain the business processing instructions and virtual asset associated data for the aircraft sent by the terminal device 102a, and input the business processing instructions and the virtual asset associated data of the aircraft into the target business processing model. Among them, the aircraft is composed of at least two components. It can be understood that the aircraft can be a transportation equipment for navigation (for example, an airplane, etc.), or it can be related components that constitute the transportation equipment. For example, when the aircraft is an airplane, the related components that constitute the airplane may include an engine, a fuselage, a landing gear, etc., and the engine, fuselage, landing gear and other components can all be used as the aircraft of this application. The business server 101 can obtain the asset data classification rules, and based on the N business types and at least two components in the asset data classification rules, divide the virtual asset related data of the aircraft to obtain S unit virtual assets, and determine the binary classification information corresponding to each unit virtual asset; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers.

[0103] The business server 101 can obtain the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model based on the binary classification information corresponding to the S unit virtual assets. The weight model parameter set includes H weight model parameters that respectively represent different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer. The specific value of H is determined by the number of components constituting the aircraft and the number of N types of business types. For example, if the N types of business types are 3 types of business, and the components constituting the aircraft include component 1 and component 2, that is, the number of components is 2, then the value of H can be equal to 6.

[0104] The business server 101 can combine the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets to obtain the fused feature vectors corresponding to the S unit virtual assets. Among them, a fused feature vector is composed of a data feature vector and a weight model parameter. For example, the S unit virtual assets include unit virtual assets a, unit virtual assets b, and unit virtual assets c, and the data feature vectors corresponding to the S unit virtual assets include data feature vectors 1a, data feature vectors 1b, and data feature vectors 1c, and the weight model parameters corresponding to the S unit virtual assets include weight model parameters 2a, weight model parameters 2b, and weight model parameters 2c. The data feature vector 1a and the weight model parameter 2a correspond to the unit virtual asset a, the data feature vector 1b and the weight model parameter 2b correspond to the unit virtual asset b, and the data feature vector 1c and the weight model parameter 2c correspond to the unit virtual asset c. The business server 101 can combine the data feature vector 1a and the weight model parameter 2a to obtain the fused feature vector corresponding to the unit virtual asset a; can combine the data feature vector 1b and the weight model parameter 2b to obtain the fused feature vector corresponding to the unit virtual asset b; and can combine the data feature vector 1c and the weight model parameter 2c to obtain the fused feature vector corresponding to the unit virtual asset c.

[0105] The business server 101 may generate a prompt text for indicating the type of identification requirement based on the business processing instruction, determine the target processing network corresponding to the type of identification requirement from the target business processing model based on the prompt text, and perform feature processing on the S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result. For example, the business processing instruction may be "Please predict the price of XX aircraft through the data I uploaded", and the prompt text may represent the key semantic information in the business processing instruction, for example, "Predict the price of XX aircraft". And the type of identification requirement indicated by the prompt text may be a predicted price requirement, and further based on the predicted price requirement, a target processing network corresponding to the predicted price requirement may be selected from multiple processing networks in the target business processing model. Further, after performing feature processing on the S fused feature vectors and the prompt text through the target processing network to obtain the feature processing result, the business server 101 may classify and identify the feature processing result to obtain a data identification result for responding to the business processing instruction. For example, when the business processing instruction is "Please predict the price of XX aircraft through the data I uploaded", the data identification result may be a predicted price for XX aircraft.

[0106] Through the above process, the asset data classification rules and business processing instructions are used to implement targeted processing of virtual asset-related data, which can improve business processing efficiency and save time; the virtual asset-related data is divided according to the asset data classification rules, and the business type and component (binary classification information) to which the unit virtual asset belongs can be obtained; the weight information (weight model parameters) of the unit virtual asset binary classification information is fully utilized, and the weight model parameters can be used to better find important data suitable for the identification requirement type, thereby increasing the influence ratio of important data, being able to better cope with business processing problems in complex data situations, and improving the recognition accuracy of the final data recognition results.

[0107] For details, see Figure 2 , Figure 2 Schematic diagram of a data processing method provided in an embodiment of the present application. Figure 2 As shown, the business server 101 can input the acquired business processing instructions and the virtual asset associated data for the aircraft into the target business processing model, wherein the aircraft is composed of at least two components. The asset data classification rule can be obtained, and based on the N business types in the asset data classification rule and the at least two components of the aircraft, the virtual asset associated data of the aircraft is divided into data to obtain S unit virtual assets (including unit virtual asset A1, unit virtual asset A2, ..., unit virtual asset A S ). And further determine the binary classification information corresponding to each unit of virtual assets, and obtain S binary classification information (including binary classification information B1, binary classification information B2, ..., binary classification information B S ). The binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers. For example, the unit virtual asset A1 belongs to business type 3 among N business types (N business types include business type 1, business type 2, and business type 3), and belongs to component 2 that constitutes the aircraft (including component 1 and component 2), then the binary classification information B1 corresponding to the unit virtual asset A1 is (business type 3, component 2). The business server 101 can obtain the weight model parameters (i.e., S weight model parameters) corresponding to the S unit virtual assets respectively from the weight model parameter set in the target business processing model based on the binary classification information corresponding to the S unit virtual assets respectively, and obtain the weight model parameter D1, obtain the weight model parameter D2, ..., obtain the weight model parameter D S. Among them, the weight model parameter set includes H weight model parameters that respectively characterize different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer. The weight model parameter set may include the weight model parameter corresponding to the binary classification information (business type 1, component 1), the weight model parameter corresponding to the binary classification information (business type 1, component 2), the weight model parameter corresponding to the binary classification information (business type 2, component 1), the weight model parameter corresponding to the binary classification information (business type 2, component 2), the weight model parameter corresponding to the binary classification information (business type 3, component 1), and the weight model parameter corresponding to the binary classification information (business type 3, component 2).

[0108] Furthermore, the business server 101 can combine the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets to obtain the fused feature vectors corresponding to the S unit virtual assets. A fused feature vector is composed of a data feature vector and a weight model parameter. That is, the business server 101 combines the data feature vector F1 corresponding to the unit virtual asset A1 to obtain the fused feature vector E1; combines the data feature vector F2 corresponding to the unit virtual asset A2 to obtain the fused feature vector E2, and so on, until the unit virtual asset A is obtained. S The corresponding data feature vector F S Combine to get the fused feature vector E S , S fused feature vectors are obtained. The business server 101 can generate a prompt text indicating the identification requirement type based on the business processing instruction, determine the target processing network corresponding to the identification requirement type from the target business processing model based on the prompt text, and perform feature processing on the S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result. Among them, the prompt text can be the key semantic information in the business processing instruction. The business server 101 can classify and identify the feature processing result to obtain a data recognition result for responding to the business processing instruction.

[0109] Through the above process, the asset data classification rules and business processing instructions are used to implement targeted processing of virtual asset-related data, which can improve business processing efficiency and save time; the virtual asset-related data is divided according to the asset data classification rules, and the business type and component (binary classification information) to which the unit virtual asset belongs can be obtained; the weight information (weight model parameters) of the unit virtual asset binary classification information is fully utilized, and the weight model parameters can be used to better find important data suitable for the identification requirement type, thereby increasing the influence ratio of important data, being able to better cope with business processing problems in complex data situations, and improving the recognition accuracy of the final data recognition results.

[0110] It is understandable that the terminal device mentioned in the embodiments of the present application may also be a computer device, and the computer device in the embodiments of the present application includes but is not limited to a terminal device or a server. In other words, the computer device may be a server or a terminal device, or it may be a system composed of a server and a terminal device. Among them, the terminal device mentioned above may be an electronic device, including but not limited to mobile phones, tablet computers, desktop computers, laptop computers, PDAs, vehicle-mounted devices, augmented reality / virtual reality (AR / VR) devices, helmet displays, smart TVs, wearable devices, smart speakers, digital cameras, cameras and other mobile Internet devices (mobile internet device, MID) with network access capabilities, or terminal devices in scenarios such as trains, ships, and flights. As Figure 1 As shown in , the terminal device may be a mobile phone (as shown in terminal device 102a), a desktop computer (as shown in terminal device 102b), a tablet computer (as shown in terminal device 102c) or a laptop computer (as shown in terminal device 102n), etc. Figure 1 Only some of the devices are listed. The servers mentioned above can be independent physical servers, server clusters or distributed systems composed of multiple physical servers, or cloud servers that provide 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, vehicle-road collaboration, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0111] For further information, see Figure 3 , Figure 3 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 1 The data processing method may be executed by a computer device, which may be Figure 1 The business server 101 shown or any terminal device in the terminal device cluster. The following will take the data processing method executed by a computer device as an example for explanation. The data processing method may at least include the following steps S301-S305:

[0112] Step S301, obtaining the business processing instruction sent by the first business object and the virtual asset associated data for the aircraft, and inputting the business processing instruction and the virtual asset associated data of the aircraft into the target business processing model; the aircraft is composed of at least two components.

[0113] In an embodiment of the present application, a computer device can obtain a business processing instruction sent by a first business object and virtual asset associated data for an aircraft, wherein the aircraft is composed of at least two components. Taking the aircraft as an engine as an example, the virtual asset associated data of the aircraft may include usage data of the components of the engine (for example, navigation-related data such as the flight duration, navigation area, etc. when the components are installed in the engine and the engine is installed in the aircraft), accident certification data of the components of the engine (for example, certification data of accidents such as fire, water ingress, and falling of the components), and asset depreciation data of the components of the engine (for example, different navigation discount rates for the navigation area when the components are installed in the engine and the engine is installed in the aircraft, because the worse the environment, the more it affects the performance of the engine, so different navigation areas will exist). Different flight discount rates), historical maintenance data of engine components (such as component maintenance time, maintenance location and recovery performance ratio after maintenance), property certificate data of engine components (such as factory license certificate, production license certificate, manufacturer, component flight permit certificate, etc.), transaction data of engine components (for example, component leasing related data: such as component leasing time, leasing amount and component return status after leasing; for example, component buying and selling related data: such as component buying and selling amount, buying and selling time, etc.; for example, component modification related data: such as component modification cost, modification time, etc.). The virtual asset related data of engine components (including usage data, accident proof data, asset depreciation data, historical maintenance data, property certificate data, transaction data, etc.) together constitute the virtual asset related data of the engine. It is understandable that the computer device can use the Real World Asset Tokenization (RWA) method to convert the virtual asset-related data of the first business object for the aircraft into a trusted asset certificate (Tokenization), and link the trusted asset certificate to the blockchain, so that when the first business object needs to transfer the virtual asset-related data for the aircraft to the second business object, the property right transfer for the aircraft can be achieved by transferring the trusted asset certificate on the blockchain. Among them, the property right transfer can be understood as the transfer of the rights attached to the item.

[0114] The computer device can input the business processing request and the virtual asset associated data of the aircraft into the target business processing model, wherein the target business processing model is used to perform a corresponding business processing process on the virtual asset associated data of the aircraft through the business processing request, thereby obtaining a result in response to the business processing request.

[0115] Step S302, obtain the asset data classification rules, and divide the virtual asset related data of the aircraft based on the N business types and at least two components in the asset data classification rules to obtain S unit virtual assets, and determine the binary classification information corresponding to each unit virtual asset; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers.

[0116] In an embodiment of the present application, a computer device may obtain an asset data classification rule, wherein N types of business types may be asset business data types, basic business data types, and transaction business data types, respectively. Taking an aircraft as an airplane, and the two components of the airplane as an engine and a landing gear as an example, the computer device divides the virtual asset associated data of the aircraft based on the N types of business types and at least two components in the asset data classification rule, and the process of obtaining S units of virtual assets may be: the computer device divides the virtual asset associated data of the aircraft into three types of data based on the asset business data type, the basic business data type, and the transaction business data type, and further divides the three types of data based on the engine and the landing gear. For example, in the classification data corresponding to the asset business data type, there are virtual asset associated data 1, virtual asset associated data 2, and virtual asset associated data 3, and the virtual asset associated data 1 belongs to the landing gear, and the virtual asset associated data 2 and the virtual asset associated data 3 belong to the engine, then the computer device divides the virtual asset associated data 1 into the landing gear under the asset business data type, and divides the virtual asset associated data 2 and the virtual asset associated data 3 into the engine under the asset business data type, thereby obtaining three units of virtual assets under the asset business data type. It should be noted that the unit virtual asset can be the virtual asset associated data itself, or it can be a tag sequence (i.e., a token tag sequence) obtained after the virtual asset associated data is segmented. It is understandable that the computer device can perform word segmentation on each virtual asset associated data in advance to obtain a corresponding tag sequence (there is a unique mapping relationship between the virtual asset associated data and the tag sequence), and store the tag sequence in a blockchain or a database. After the computer device divides the virtual asset associated data, the tag sequence corresponding to the virtual asset associated data can be directly obtained in the database or blockchain as a unit virtual asset; it can also be directly segmented on the virtual asset associated data after the division when the virtual asset associated data is divided, to obtain the corresponding tag sequence, and determine the tag sequence as a unit virtual asset. Further, the computer device can obtain the business type and component to which each unit virtual asset belongs, and combine the business type and component into a binary classification information of the unit virtual asset. For example, the binary classification information of the virtual asset associated data 1 mentioned above is (asset business data type, landing gear), the binary classification information of the virtual asset associated data 2 is (asset business data type, engine), and the binary classification information of the virtual asset associated data 3 is (asset business data type, engine).

[0117] Optionally, the virtual asset-related data can be associated with the corresponding RWA data, that is, after the computer device divides the virtual asset-related data of aviation by N business types and at least two components, the trusted asset certificates of different unit asset data can be classified and stored on the blockchain according to the N business types and components. The trusted asset certificate belongs to RWA data, which can realize asset transfer or property transfer on the blockchain. For example, the trusted asset certificate can be transferred to another business object to realize the transfer of the asset data corresponding to the trusted asset certificate to another business object. The virtual asset-related data sent by the first business object can carry the trusted asset certificate. The computer device can detect whether the trusted asset certificate carried by the virtual asset-related data exists in the block on the blockchain. If the computer device detects that the trusted asset certificate exists in the corresponding block, the virtual asset-related data corresponding to the trusted asset certificate can be determined as unit asset data, thereby obtaining the divided S unit virtual assets.

[0118] Step S303, based on the binary classification information corresponding to the S unit virtual assets, obtain the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model; the weight model parameter set includes H weight model parameters that respectively represent different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer.

[0119] In an embodiment of the present application, the computer device can determine and obtain the weight model parameters corresponding to the binary classification information corresponding to S units of virtual assets from the weight model parameters based on the mapping relationship between the weight model parameters and the binary classification information. Among them, the weight model parameter set can be obtained by the computer device when training the initial business processing model, and the weight model parameter set includes H weight model parameters that respectively represent different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer.

[0120] Step S304, combining the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets to obtain fused feature vectors corresponding to the S unit virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter.

[0121] In an embodiment of the present application, the target business processing model may include a multimodal feature extraction component (e.g., Contrastive Language-Image Pre-training, CLIP), and the computer device may extract features of S unit virtual assets respectively through the multimodal feature extraction component to obtain data feature vectors corresponding to the S unit virtual assets. CLIP can not only process text, but also process images, videos and other formats of content, which can help the target business processing model quickly determine the data feature vectors corresponding to different types of unit virtual assets. The computer device can determine the degree of association between the S unit virtual assets and the identification requirement type indicated by the business processing instruction. Among them, the degree of association is used to characterize the correlation between the two. The computer device can obtain the degree of association between the unit virtual asset and the identification requirement type by calculating the attention score, vector similarity or Pearson correlation coefficient between the unit virtual asset and the identification requirement type, which is not limited here. Further, the computer device can assign different data association weights to the S unit virtual assets based on different degrees of association. For example, linear allocation or exponential allocation based on the degree of association can be adopted. It can be understood that the greater the degree of association, the greater the data association weight. The computer device can combine the data feature vectors corresponding to the S unit virtual assets, the data association weights corresponding to the S unit virtual assets, and the weight model parameters corresponding to the S unit virtual assets to obtain the fused feature vectors corresponding to the S unit virtual assets. Among them, a fused feature vector is composed of a data feature vector, a data association weight, and a weight model parameter. For example, there is a unit virtual asset x, and the fused feature vector corresponding to the unit virtual asset x is composed of the data feature vector corresponding to the unit virtual asset x, the data association weight, and the weight model parameter.

[0122] Step S305, based on the business processing instruction, a prompt text is generated to indicate the identification requirement type; based on the prompt text, a target processing network corresponding to the identification requirement type is determined from the target business processing model; feature processing is performed on the S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result; the feature processing result is classified and identified to obtain a data recognition result for responding to the business processing instruction.

[0123] In an embodiment of the present application, a computer device may call a natural language processing component, analyze the business processing instruction through the natural language processing component, and obtain a prompt text for indicating the type of identification requirement. For example, the business processing instruction may be "Please predict the price of XX aircraft through the data I uploaded", and the prompt text may be "Predict the price of XX aircraft". And the type of identification requirement indicated by the prompt text may be a price prediction requirement. Further, the computer device may determine the target processing network corresponding to the identification requirement type from the target business processing model based on the prompt text. It can be understood that the target business processing model includes multiple business processing networks, and the computer device may select the required business processing network from the multiple business processing networks as the target processing network based on the identification requirement type, and perform feature processing on S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result. The feature processing result may further be classified and identified to obtain a data identification result for responding to the business processing instruction.

[0124] Through the above process, through the target business processing model, using the asset data classification rules and at least two components to divide the virtual asset related data, the business type and component (two-tuple classification information) to which the unit virtual asset belongs can be obtained, and there is no need for manual participation in data division, saving labor costs; the weight information (weight model parameters) of the unit virtual asset two-tuple classification information is fully utilized, and the weight model parameters can be used to better find important data suitable for the identification requirement type, thereby increasing the influence ratio of important data, and being able to better deal with business processing problems in complex data situations, and improve the recognition accuracy of the final data recognition results. And based on the prompt text generated by the business processing instruction, the target processing network can be determined from the target business processing model to achieve targeted processing of the unit virtual assets, which can improve business processing efficiency and save time.

[0125] For further information, see Figure 4 , Figure 4 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 2 The data processing method can be executed by a computer device, which can be Figure 1 The business server 101 shown or any terminal device in the terminal device cluster. The following will take the data processing method executed by a computer device as an example for description. The data processing method may at least include the following steps S401-S406:

[0126] Step S401, obtaining the business processing instruction sent by the first business object and the virtual asset associated data for the aircraft, and inputting the business processing instruction and the virtual asset associated data of the aircraft into the target business processing model; the aircraft is composed of at least two components.

[0127] In the embodiment of the present application, the specific implementation process of step S401 can be referred to as follows. Figure 3 The specific description of step S301 shown in FIG. 1 is not repeated here.

[0128] Step S402, obtain the asset data classification rules, and divide the virtual asset related data of the aircraft based on the N business types and at least two components in the asset data classification rules to obtain S unit virtual assets, and determine the binary classification information corresponding to each unit virtual asset; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers.

[0129] In the embodiment of the present application, the computer device can obtain the asset data classification rule, and classify the virtual asset associated data of the aircraft through the N business types in the asset data classification rule to obtain M types of classified data; the business types corresponding to the M types of classified data belong to the N business types, and M is a positive integer less than or equal to N. Specifically, the N types of business types include asset business data types, basic business data types, and transaction business data types; if M is equal to N, the M types of classified data include first classified data, second classified data, and third classified data. The computer device can obtain the data key fields corresponding to the asset business data type; wherein the data key fields include the usage record key field, the maintenance record key field and the depreciation record key field; subsequently, the historical usage records corresponding to the usage record key field, the historical maintenance records corresponding to the maintenance record key field and the historical depreciation records corresponding to the depreciation record key field can be obtained in the virtual asset associated data of the aircraft, and the historical usage records, historical maintenance records and historical depreciation records can be determined as the first classification data corresponding to the asset business data type; the factory key field corresponding to the basic business data type can be obtained, and the factory associated data corresponding to the factory key field can be obtained in the virtual asset associated data of the aircraft, and the factory associated data can be determined as the second classification data corresponding to the basic business data type; the transaction key field corresponding to the transaction business data type can be obtained, and the transaction associated data corresponding to the transaction key field can be obtained in the virtual asset associated data of the aircraft, and the transaction associated data can be determined as the third classification data corresponding to the transaction business data type.

[0130] It should be understood that each business type may correspond to different key fields, and each key field may correspond to a type of data. Then, data belonging to different business types may be obtained in the aircraft's virtual asset-related data according to different key fields. For example, the asset business data type may correspond to the usage record key field, the maintenance record key field, and the depreciation record key field. Then, according to the usage record key field, the historical usage record (such as the historical usage record of each component of the aircraft) may be obtained in the aircraft's virtual asset-related data. According to the maintenance record key field, the historical maintenance record (such as the historical maintenance record of each component of the aircraft) may be obtained in the aircraft's virtual asset-related data. According to the depreciation record key field, the historical depreciation record (such as the historical accident record of each component of the aircraft) may be obtained in the aircraft's virtual asset-related data. And the historical usage record, historical maintenance record, and historical depreciation record may all be used as data corresponding to the asset business type. If the basic business data type can correspond to the factory key field, then the factory-related data (such as factory permit certificate, production permit certificate, etc.) can be obtained in the aircraft's virtual asset-related data based on the factory key field; if the transaction business data type can correspond to the transaction key field, then the transaction-related data (such as rent, lease term, return status after lease, etc.) can be obtained in the aircraft's virtual asset-related data based on the transaction key field.

[0131] Further, the computer device can divide each type of classified data into components based on at least two components to obtain S unit virtual assets. One unit virtual asset corresponds to one business type and one component. Figure 5 , Figure 5 is a data classification diagram provided in an embodiment of the present application. Figure 5 As shown, the terminal device 501 may be the terminal device 102a mentioned above, and the object a may be the first business object mentioned above. The object a may send the virtual asset associated data 502 of the aircraft (composed of component 1 and component 2) to the computer device 503, wherein the virtual asset associated data 502 of the aircraft may include virtual asset associated data 1, virtual asset associated data 2, virtual asset associated data 3, virtual asset associated data 4, virtual asset associated data 5, and virtual asset associated data 6. After receiving the virtual asset associated data 502 of the aircraft, the computer device 503 may obtain an asset data classification rule, which includes N business types (including business type 1, business type 2, and business type 3) for data classification; further, the virtual asset associated data 502 of the aircraft may be classified according to the business type in the asset data classification rule. Figure 5As shown, in the virtual asset association data 502 (including virtual asset association data 1, virtual asset association data 2, virtual asset association data 3, virtual asset association data 4, virtual asset association data 5 and virtual asset association data 6) of the aircraft, it is possible to obtain virtual asset association data whose business type is the business type 1 (including virtual asset association data 1 and virtual asset association data 3), obtain virtual asset association data whose business type is the business type 2 (including virtual asset association data 2 and virtual asset association data 4), and obtain virtual asset association data whose business type is the business type 3 (including virtual asset association data 5 and virtual asset association data 6).

[0132] like Figure 5 As shown, the virtual asset associated data under one business type can be used as a classification data. Then, according to business type 1, business type 2 and business type 3, the virtual asset associated data of the item can be divided into three classification data. The three classification data can include classification data under business type 1 (including virtual asset associated data 1 and virtual asset associated data 3), classification data under business type 2 (including virtual asset associated data 2 and virtual asset associated data 4), and classification data under business type 3 (including virtual asset associated data 5 and virtual asset associated data 6). Furthermore, the computer device can divide each classification data into components according to the components of the aircraft (including component 1 and component 2). For example, for the classified data under business type 1, since virtual asset associated data 1 is the virtual asset associated data of component 1 and virtual asset associated data 3 is the virtual asset associated data of component 2, the classified data under business type 1 can be further divided into two types of data (one is virtual asset associated data 1 belonging to component 1, and the other is virtual asset associated data 3 belonging to component 2); similarly, for the classified data under business type 2, since virtual asset associated data 2 and virtual asset associated data 4 are both virtual asset associated data of component 2, the classified data under business type 2 can be further divided into one type of data (virtual asset associated data 2 and virtual asset associated data 4 belonging to component 2); similarly, for the classified data under business type 3, since virtual asset associated data 5 and virtual asset associated data 6 are both virtual asset associated data of component 1, the classified data under business type 3 can be further divided into one type of data (virtual asset associated data 5 and virtual asset associated data 6 belonging to component 1).

[0133] At this time, the 6 virtual asset-related data are divided into 4 types of data (including 6 unit virtual assets) according to the business type and components. The computer device can determine the 6 virtual asset-related data as 6 unit virtual asset data, or it can perform word segmentation on the 6 virtual asset-related data to obtain 6 token marking sequences, and determine the 6 token marking sequences as 6 unit virtual assets. Furthermore, the computer device can obtain the business types and components corresponding to the S unit virtual assets respectively, combine the business types and components corresponding to the S unit virtual assets, and obtain the binary classification information corresponding to the S unit virtual assets. For example, Figure 5 The binary classification information corresponding to the unit virtual asset corresponding to the virtual asset association data 1 shown is (business type 1, component 1).

[0134] Step S403, based on the binary classification information corresponding to the S unit virtual assets, obtain the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model; the weight model parameter set includes H weight model parameters that respectively represent different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer.

[0135] In the embodiment of the present application, the binary classification information corresponding to the S units of virtual assets includes binary classification information B i , i is a positive integer less than or equal to S; the computer device can obtain the weight model parameter set and parameter matching rules in the target business processing model; wherein the parameter matching rule is used to indicate the mapping relationship between the weight model parameters in the weight model parameter set and the binary classification information; for example, when the N types of business types include business type 1, business type 2, and business type 3, and the parts constituting the aircraft include part 1 and part 2, the parameter matching rules can be as shown in Table 1:

[0136] Table 1

[0137] Bigram classification information Weight model parameters (Business Type 1, Component 1) Parameter G1 (Business Type 1, Component 2) Parameter G2 (Business Type 2, Component 1) Parameter G3 (Business Type 2, Component 2) Parameter G4 (Business type 3, component 1) Parameter G5 (Business type 3, component 2) Parameter G6

[0138] It should be noted that the weight model parameter can be a vector, and the computer device can determine the binary classification information set based on the weight model parameter set and the parameter matching rule; wherein the binary classification information set can include “(business type 1, component 1), (business type 1, component 2), (business type 2, component 1), (business type 2, component 2), (business type 3, component 1), (business type 3, component 2)”. The computer device can combine the binary classification information set with the binary classification information B iThe matched binary classification information is determined as the target binary classification information. Based on the target binary classification information and the parameter matching rule, the target weight model parameter corresponding to the target binary classification information can be obtained from the weight model parameter, and the target weight model parameter can be determined as the binary classification information B i The corresponding weight model parameters. For example, the binary classification information B i When it is (business type 2, component 2), the corresponding weight model parameter can be parameter G4. Among them, the weight model parameter set includes H weight model parameters that respectively characterize different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer. The value of H is determined by the N business types and the component data that make up the aircraft. For example, if N is 3 and the number of components that make up the aircraft is 5, then H can be 15. It can be understood that the binary classification information set corresponding to the weight model parameter set includes but is not limited to the binary classification information corresponding to the S unit virtual assets obtained after the virtual asset association data of the aircraft sent by the first business object is divided. For example, the binary classification information set may include "(business type 1, component 1), (business type 1, component 2), (business type 2, component 1), (business type 2, component 2), (business type 3, component 1), (business type 3, component 2)". However, the binary classification information corresponding to the S unit virtual assets may only include "(business type 1, component 1), (business type 2, component 1), (business type 2, component 2), (business type 3, component 1)".

[0139] Step S404, combining the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets to obtain fused feature vectors corresponding to the S unit virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter.

[0140] In the embodiment of the present application, the S unit virtual assets include unit virtual assets A i , the data feature vectors corresponding to the S unit virtual assets include unit virtual asset A i The corresponding data feature vector C i , the weight model parameters corresponding to the S units of virtual assets include unit virtual asset A i The corresponding weight model parameter D i , i is a positive integer less than or equal to S. The computer device can determine the identification requirement type and the unit virtual asset A indicated by the business processing instruction. i The correlation between virtual assets A based on the correlation iAssign data association weights. The association between the unit virtual asset and the identification requirement type can be obtained by calculating the attention score, vector similarity, or Pearson correlation coefficient between the unit virtual asset and the identification requirement type, and there is no limitation here. The computer device can calculate the data feature vector C i and weight model parameter D i Splice and obtain data splicing features, and multiply the data splicing features and the data association weight to determine the unit virtual asset A i The corresponding fusion feature vector. A possible concatenation method can be found in formula ①:

[0141] F=t*(C i +D i ) ①

[0142] As shown in formula ①, F is used to represent the unit virtual asset A i The corresponding fusion feature vector, t, is used to represent the unit virtual asset A i The corresponding data association weight.

[0143] Step S405, based on the business processing instruction, a prompt text is generated to indicate the identification requirement type; based on the prompt text, a target processing network corresponding to the identification requirement type is determined from the target business processing model; feature processing is performed on the S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result; the feature processing result is classified and identified to obtain a data recognition result for responding to the business processing instruction.

[0144] In an embodiment of the present application, a computer device may generate a prompt text for indicating the type of identification demand based on a business processing instruction through a natural language processing technology. The computer device may determine a target processing network corresponding to the type of identification demand from a target business processing model based on the prompt text. For example, when the prompt text is "predict the price of XX" or text related to the predicted price, the target processing network may be a pricing prediction network. At this time, the computer device performs feature processing on S fused feature vectors and the prompt text through the target processing network, and the specific implementation process of obtaining the feature processing result may be: the computer device extracts features from the prompt text to obtain prompt text features. Specifically, the target business processing model also includes a text encoding component, and the computer device may perform text segmentation on the prompt text through the text encoding component to obtain a word-unit sequence corresponding to the prompt text. Each word-unit in the word-unit sequence corresponding to the prompt text is encoded and processed to obtain prompt text features. Among them, a word-unit in the word-unit sequence refers to the smallest basic unit obtained after the prompt text is segmented. The method of text segmentation may be based on word segmentation (word-based), character segmentation (character-based), or subword segmentation (subword-based), and the embodiment of the present application is not limited here. Further, the computer device can obtain the environmental factor information of the object to be priced indicated by the prompt text based on the search engine associated with the pricing prediction network. Among them, the object to be priced refers to the main object in the prompt text. For example, if the prompt text is "predict the rental price of the XS aircraft", the object to be priced is "XS aircraft". Environmental factor information can refer to various related factors that can be retrieved based on the search engine that cause the rental price of the XS aircraft. Environmental factor information can include external market environment factors, the popularity of the XS aircraft, the credit status of the lessee of the XS aircraft, and the current ESG {Environment, Society, Governance} standards. Since the environmental factor information may include relevant data such as text, pictures and videos, the computer device can extract features of the environmental factor information through a multimodal feature extraction component to obtain environmental factor features. Further, the environmental factor features, the prompt text features and S fused feature vectors can be fused to obtain feature processing results. Common fusion methods include vector concatenation or weighted average, which are not limited here.

[0145] Furthermore, the computer device classifies and identifies the feature processing results, and the specific implementation process of obtaining the data identification results for responding to the business processing instructions can be: the computer device can obtain P price range labels corresponding to the pricing prediction network, and the estimated market information corresponding to the P price range labels, where P is a positive integer. The price range label can be set according to the historical pricing data of the object to be priced, and a price range label is used to indicate a price range. The estimated market information is used to indicate various pricing factor data of the corresponding price range. The more relevant a feature processing result is to an estimated market information, the more relevant the predicted price of the object to be priced is to the price range associated with the estimated market information. The computer device can extract features from the P estimated market information respectively to obtain P price market features. The feature processing results and the P price market features can be cross-attention processed to obtain P first cross-attention scores. Specifically, the computer device can combine the P price market features by vector splicing to obtain a price combination feature. The feature processing result can be used as the query vector in the cross-attention function, the price combination feature can be used as the key vector in the cross-attention function, and the price combination feature can be used as the value vector in the cross-attention function; the product of the query vector and the key vector can be determined as the first fused feature through the cross-attention function; the number of dimensions corresponding to the price combination feature can be obtained, and the product of the first fused feature and the inverse of the number of dimensions can be determined as the second fused feature; based on the activation subfunction in the cross-attention function, the second fused feature can be converted into the first activated feature; the product of the first activated feature and the value vector can be used to determine the attention processing sequence; the attention processing sequence is composed of P first cross-attention scores. A possible cross-attention function can be found in formula ②:

[0146]

[0147] As shown in formula ②, Attention(Q,K,V) represents the cross attention function, Q is used to represent the query vector query, K is used to represent the key vector key, and V is used to represent the value vector value; T represents transposition, K T represents the transposed matrix of K, d k Used to represent the number of dimensions corresponding to the price combination feature; is the first activated feature, Processing sequence for attention.

[0148] The computer device may determine the price range label corresponding to the estimated market information associated with the first target cross-attention score as the data identification result for responding to the business processing instruction. The first target cross-attention score is the largest first cross-attention score among the P first cross-attention scores. That is, the price range label most associated with the feature processing result is obtained from the P price range labels, and the price range label is determined as the data identification result for responding to the business processing instruction.

[0149] Optionally, when the prompt text is "analyzing the health status of the lease portfolio", "analyzing the operating efficiency of the lease portfolio" or "analyzing the financial performance of virtual asset-related data", etc., and the target processing network determined based on the prompt text is a data analysis network, the computer device performs feature processing on the S fused feature vectors and the prompt text through the target processing network, and the specific implementation process of obtaining the feature processing result can be: the computer device can extract the features of the prompt text through the text encoding component to obtain the prompt text features; and fuse the prompt text features and the S fused feature vectors through vector splicing or weighted average to obtain the feature processing result. Further, the computer device classifies and identifies the feature processing result, and the specific implementation process of obtaining the data identification result for responding to the business processing instruction can be: the computer device can obtain Q data classification labels corresponding to the data analysis network, and the label association information corresponding to the Q data classification labels. The data classification labels can be "high, medium, low", "excellent, good, poor" and other classification labels used to analyze the health status of the lease portfolio, analyze the operating efficiency of the lease portfolio or analyze the financial performance of virtual asset-related data. The tag association information is used to indicate the data associated with the corresponding data classification tag for comparison with the unit virtual asset, that is, the more relevant the unit virtual asset is to a tag association information, the more relevant the unit virtual asset is to the data classification tag corresponding to the tag association information.

[0150] The computer device can perform feature extraction on the Q tag association information respectively to obtain Q data classification features; Q is a positive integer. The feature processing results and the Q data classification features can be cross-attention processed to obtain Q second cross-attention scores. The specific implementation process can refer to the specific description process of obtaining the P first cross-attention scores mentioned above, which will not be repeated here. The computer device can determine the data classification label corresponding to the tag association information associated with the second target cross-attention score as the data recognition result for responding to the business processing instruction; wherein the second target cross-attention score is the largest second cross-attention score among the Q second cross-attention scores. That is, through cross-attention processing, the tag association information most relevant to the unit virtual asset and the prompt text can be obtained.

[0151] Step S406: perform data processing on the data recognition result and the prompt text to obtain a business result for the business processing instruction.

[0152] In an embodiment of the present application, the computer device can perform information retrieval on the data recognition result and the prompt text through a large language model to obtain an information retrieval result. A large language model (LLM) is a model trained based on a large amount of text data, which can generate coherent text and perform multiple language processing tasks. The computer device can obtain a template library associated with the target processing network, and obtain a business template corresponding to the prompt text from the template library; wherein, when the target processing network is a pricing prediction network, the template library may include a business template that analyzes or recommends the predicted price to ensure that the final pricing can be used to maximize revenue and ensure the competitiveness of the final pricing. When the target processing network is a data analysis network, the template library may include an insight report template for financial performance, lease portfolio health and operational efficiency, or a lease portfolio optimization suggestion template. The computer device can determine the text to be updated in the business template, replace the text to be updated in the business template based on the data recognition result and the information retrieval result, obtain the updated business text, and determine the updated business text as the business result for the business processing instruction. For example, the first business object can learn the leasing performance of a certain leasing combination through the business result, and decide whether to change the leasing combination to obtain the final leasing combination. The leasing combination refers to a combination of several aircraft or aircraft components that are leased together to obtain the best leasing price.

[0153] Optionally, if the recognition requirement type indicated by the prompt text includes a drawing processing requirement, the computer device can identify the drawing image type according to the drawing processing requirement, and obtain an image template with the drawing image type in the image template library. The image template library includes multiple types of image modules, such as statistical graphs, histograms, line graphs, and other image templates for mathematical statistical analysis. A Gaussian noise image is generated according to the image template and the initial noise data; the data recognition result and the prompt text can be input into the large language model, and the data recognition result and the prompt text are fused and expanded by the large language model to obtain the drawing instruction text. The Gaussian noise image and the drawing instruction text can be input into the Venn diagram model, and the Gaussian noise image is feature extracted by the Venn diagram model to obtain Gaussian noise features. The latent variable distribution can be obtained in the forward diffusion network layer of the Venn diagram model, wherein the latent variable distribution is a concept distribution with added noise, such as a Gaussian distribution. The computer device can continuously add random noise vectors to the Gaussian noise features in T time steps to obtain a forward noise vector. Furthermore, the drawing instruction text can be feature-encoded by the Wensheng graph model to obtain drawing text features, and the Gaussian noise image can be denoised according to the forward noise vector and the drawing text features to obtain a drawing data graph; optionally, if there is associated historical reference data for the prompt text and the data recognition result, for example, when the data recognition result is a predicted price range, there is historical reference data such as historical predicted prices or historical rental prices for the object to be priced, the historical reference data can be input into the Wensheng graph model, and the Gaussian noise image can be denoised by the reference data features, forward noise vectors and drawing text features of the historical reference data to obtain a drawing data graph. The computer device can determine the data recognition result and the drawing data graph as the business result for the business processing instruction. For example, the first business object can determine the final price of the object to be priced according to the data recognition result and the drawing data graph.

[0154] Through the above process, through the target business processing model, the virtual asset associated data is divided using the asset data classification rules, and the business type and component (binary classification information) to which the unit virtual asset belongs can be obtained, and there is no need for manual participation in data division, saving labor costs; the weight information (weight model parameters) of the business type or component to which the unit virtual asset belongs, as well as the data association weights directly between the unit virtual asset and the business processing instruction are fully utilized, and the important data suitable for the identification demand type can be better found through the weight model parameters and the data association weights, thereby increasing the influence ratio of the important data, and being able to better deal with business processing problems under complex data conditions, thereby improving the accuracy of the final data recognition results. Based on the prompt text generated by the business processing instruction, the target processing network (including pricing prediction network and data analysis network, etc.) can be determined from the target business processing model to achieve targeted processing of the unit virtual asset, which can improve business processing efficiency and save time. The final business result of the response to the business processing instruction is obtained through the analysis and processing of the data recognition result, providing more professional suggestions for the first business object, which can help the first business object achieve a better solution, without the need to manually analyze various types of data, and obtain the corresponding business results, saving labor costs.

[0155] For further information, see Figure 6 , Figure 6 This is a flow diagram of a data processing method provided in an embodiment of the present application. Figure 3 The data processing method can be executed by a computer device, which can be Figure 1 The business server 101 shown or any terminal device in the terminal device cluster. The following will take the data processing method executed by a computer device as an example for explanation. The data processing method may at least include the following steps S601-S605:

[0156] Step S601, obtain sample processing instructions, virtual asset sample data for a sample aircraft and sample identification results, and input the sample processing instructions and the virtual asset sample data of the sample aircraft into an initial business processing model; the sample aircraft is composed of at least two sample components.

[0157] In the embodiment of the present application, the sample aircraft may be the same as the aircraft mentioned above, and the virtual asset sample data for the sample aircraft may include all data of at least two sample components of the sample aircraft under N types of business. The sample identification result is the data result that should be obtained by processing the virtual asset sample data of the sample aircraft under the sample processing instruction, and is used for data comparison with the sample prediction result obtained by processing the virtual asset sample data of the sample aircraft based on the sample processing instruction in the initial business processing model.

[0158] Step S602, obtain the asset data classification rules, and divide the virtual asset sample data of the sample aircraft based on the N business types and at least two sample components in the asset data classification rules to obtain R unit sample virtual assets, and determine the sample two-tuple classification information corresponding to each unit sample virtual asset; the sample two-tuple classification information is used to characterize the business type and sample component to which the unit sample virtual asset belongs; N and R are both positive integers.

[0159] In the embodiment of the present application, the specific implementation process of step S602 can be found in Figure 4 The relevant description in step S402 shown in is not repeated here.

[0160] Step S603, respectively assign random initialization model parameters to the binary classification information corresponding to the R unit sample virtual assets, and determine the R initialization model parameters as the initialization model parameter set; combine the sample feature vectors corresponding to the R unit sample virtual assets and the initialization model parameters corresponding to the R unit sample virtual assets to obtain sample fusion feature vectors corresponding to the R unit sample virtual assets; a sample fusion feature vector is composed of a sample feature vector and an initialization model parameter.

[0161] In an embodiment of the present application, a computer device can randomly generate R initialization vectors, and determine the R initialization vectors as initialization model parameters, respectively, for the binary classification information corresponding to the R unit sample virtual assets, and assign random initialization model parameters, and the R initialization model parameters can be determined as an initialization model parameter set. The computer device can determine the sample correlation between the R unit sample virtual assets and the sample identification requirement type indicated by the sample processing instruction. Different sample association weights are assigned to the R unit sample virtual assets through the sample correlation. Further, the sample feature vectors corresponding to the R unit sample virtual assets, the sample association weights corresponding to the R unit sample virtual assets, and the initialization model parameters corresponding to the R unit sample virtual assets are combined to obtain sample fusion feature vectors corresponding to the R unit sample virtual assets. One of the sample fusion feature vectors is composed of a sample feature vector, a sample association weight, and an initialization model parameter.

[0162] Step S604, based on the sample processing instruction, a sample prompt text is generated to indicate the sample identification requirement type; based on the sample prompt text, a sample processing network corresponding to the sample identification requirement type is determined from the initial business processing model; feature processing is performed on R sample fusion feature vectors and the sample prompt text through the sample processing network to obtain a sample feature processing result; the sample feature processing result is classified and identified to obtain a sample prediction result for responding to the sample processing instruction.

[0163] In an embodiment of the present application, a computer device may call a natural language processing component, and through the natural language processing component, analyze the sample processing instruction to obtain a sample prompt text indicating the sample identification requirement type. Further, the computer device may determine a sample processing network corresponding to the sample identification requirement type from the initial business processing model based on the sample prompt text. It is understandable that the initial business processing model includes multiple initial processing networks, and the computer device may select the required initial processing network from the multiple initial processing networks based on the sample identification requirement type as the sample processing network, and perform feature processing on R sample fusion feature vectors and the sample prompt text through the sample processing network to obtain a sample feature processing result. The sample feature processing result may further be classified and identified to obtain a sample prediction result for responding to the sample processing instruction.

[0164] Step S605, based on the sample prediction results and the sample identification results, a model loss value is generated, and each initialization model parameter in the initialization model parameter set and the parameters of the sample processing network are adjusted based on the model loss value until the initial business processing model meets the model convergence conditions, and the target business processing model is obtained; the target business processing model includes a weight model parameter set, and the weight model parameter set includes the weight model parameters obtained after the initialization model parameters are adjusted.

[0165] In the embodiment of the present application, the computer device can generate a model loss value based on the sample prediction result and the sample recognition result. A possible method for generating the model loss value can be seen in formula ③:

[0166]

[0167] As shown in formula ③, Cosine Loss is used to represent the cosine similarity loss value (i.e., model loss value) between the sample prediction result and the sample recognition result, and cosine_similarity is used to represent the cosine similarity between the sample prediction result and the sample recognition result. A possible cosine similarity calculation formula can be found in formula ④:

[0168]

[0169] As shown in formula ④, cosine_similarity(A,B) is used to represent the cosine similarity between vector A and vector B, that is, the computer device can vectorize the sample prediction result to obtain vector A, and can vectorize the sample recognition result to obtain vector B, and calculate the cosine similarity between vector A and vector B. Where "·" represents the dot product, A·B represents the dot product result between vector A and vector B, ‖A‖ represents the modulus length of vector A, and ‖B‖ represents the modulus length of vector B.

[0170] Furthermore, the computer device can adjust the parameters of each initialization model parameter in the initialization model parameter set and the parameters of the sample processing network based on the model loss value until the parameters converge. It should be noted that the sample data for parameter adjustment of the initial business processing model includes multiple batches of sample processing instructions, and multiple batches of sample processing instructions and sample identification results corresponding to the virtual asset sample data of the sample aircraft. A batch of sample processing instructions can correspond to one of the multiple initial processing networks in the initial business processing model. The computer device can adjust the parameters of the initialization model parameters in the initialization model parameter set and the sample processing network corresponding to each batch based on the model loss value of each batch until the initial business processing model satisfies the model convergence condition and obtains the target business processing model. Among them, the target business processing model includes a weight model parameter set, and the weight model parameter set includes the weight model parameters obtained after the initialization model parameters are adjusted.

[0171] Through the above process, the sample prediction results output by the initial business processing model are combined with the strategy of parameter adjustment through sample identification to obtain a set of weight model parameters, which can enable the weight model parameters to learn the demand relationship between sample processing instructions and sample binary classification information of virtual asset sample data, that is, the importance of important data in the virtual asset sample data used to implement sample processing instructions. Therefore, when the target business processing model processes business processing instructions and related data, the business processing instructions and related data as well as the weight model parameters of the related data can be deeply combined, thereby improving the influence ratio of important data in the related data used to implement business processing instructions, thereby improving the accuracy and stability of business processing.

[0172] For further information, see Figure 7 , Figure 7 A data processing device provided in an embodiment of the present application is shown in FIG. Figure 1 The data processing device 700 may be a computer program (including program code, etc.) running in a computer device. For example, the data processing device 700 may be an application software. The data processing device 700 may be used to execute the corresponding steps of the method provided in the embodiment of the present application. Figure 7 As shown, the data processing device 700 can be used to Figure 3 , Figure 4 and Figure 6 The computer device in the corresponding embodiment, specifically, the data processing device 700 may include: a data acquisition module 11, a data division module 12, a parameter acquisition module 13, a feature combination module 14, a network determination module 15, a classification identification module 16, and a result determination module 17.

[0173] The data acquisition module 11 is used to acquire the business processing instruction sent by the first business object and the virtual asset associated data for the aircraft, and input the business processing instruction and the virtual asset associated data of the aircraft into the target business processing model; the aircraft is composed of at least two components;

[0174] The data partitioning module 12 is used to obtain the asset data classification rules, and based on the N business types and at least two components in the asset data classification rules, the virtual asset associated data of the aircraft is partitioned to obtain S unit virtual assets, and determine the binary classification information corresponding to each unit virtual asset; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers;

[0175] The parameter acquisition module 13 is used to acquire the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model based on the binary classification information corresponding to the S unit virtual assets; the weight model parameter set includes H weight model parameters representing different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer;

[0176] The feature combination module 14 is used to combine the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets to obtain fused feature vectors corresponding to the S unit virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter;

[0177] The network determination module 15 is used to generate a prompt text indicating the identification requirement type based on the business processing instruction, determine the target processing network corresponding to the identification requirement type from the target business processing model based on the prompt text, and perform feature processing on the S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result;

[0178] The classification and identification module 16 is used to classify and identify the feature processing results to obtain data identification results for responding to the business processing instructions.

[0179] In a possible implementation, the data partitioning module 12 is used to partition the virtual asset associated data of the aircraft based on the N business types and at least two components in the asset data classification rule to obtain S unit virtual assets. When determining the binary classification information corresponding to each unit virtual asset, the data partitioning module 12 is specifically used to perform the following operations:

[0180] The virtual asset associated data of the aircraft is classified according to the N business types in the asset data classification rule to obtain M types of classified data; the business types corresponding to the M types of classified data belong to the N business types, and M is a positive integer less than or equal to N;

[0181] In each type of classified data, each type of classified data is divided into components based on at least two components to obtain S unit virtual assets; one unit virtual asset corresponds to one business type and one component;

[0182] The business types and components corresponding to the S unit virtual assets are obtained respectively, and the business types and components corresponding to the S unit virtual assets are combined to obtain the binary classification information corresponding to the S unit virtual assets respectively.

[0183] In a possible implementation, the N types of business types include asset business data types, basic business data types, and transaction business data types; if M is equal to N, the M types of classified data include first classified data, second classified data, and third classified data; the data partitioning module 12 is used to classify the virtual asset associated data of the aircraft according to the N types of business types in the asset data classification rule, and when the M types of classified data are obtained, the data partitioning module 12 is specifically used to perform the following operations:

[0184] Obtain the data key fields corresponding to the asset business data type; the data key fields include the usage record key fields, maintenance record key fields, and impairment record key fields;

[0185] In the virtual asset associated data of the aircraft, historical use records corresponding to the use record key field, historical maintenance records corresponding to the maintenance record key field, and historical damage records corresponding to the damage record key field are obtained, and the historical use records, historical maintenance records, and historical damage records are determined as first classification data corresponding to the asset business data type;

[0186] Obtain the factory key field corresponding to the basic business data type, obtain the factory associated data corresponding to the factory key field in the virtual asset associated data of the aircraft, and determine the factory associated data as the second classification data corresponding to the basic business data type;

[0187] Obtain the transaction key field corresponding to the transaction business data type, obtain the transaction related data corresponding to the transaction key field in the virtual asset related data of the aircraft, and determine the transaction related data as the third classification data corresponding to the transaction business data type.

[0188] In a possible implementation, the binary classification information corresponding to the S units of virtual assets includes binary classification information B i, i is a positive integer less than or equal to S; the parameter acquisition module 13 is used to obtain the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model based on the binary classification information corresponding to the S unit virtual assets, and the parameter acquisition module 13 is specifically used to perform the following operations:

[0189] Obtaining a weight model parameter set and a parameter matching rule in a target business processing model; the parameter matching rule is used to indicate a mapping relationship between a weight model parameter in the weight model parameter set and binary classification information;

[0190] Determine a binary classification information set based on a weight model parameter set and a parameter matching rule;

[0191] Combine the binary classification information set with the binary classification information B i The matched bigram classification information is determined as the target bigram classification information;

[0192] Based on the target binary classification information and parameter matching rules, the target weight model parameters corresponding to the target binary classification information are obtained from the weight model parameters, and the target weight model parameters are determined as the binary classification information B i The corresponding weight model parameters.

[0193] In a possible implementation, the S unit virtual assets include unit virtual asset A i , the data feature vectors corresponding to the S unit virtual assets include unit virtual asset A i The corresponding data feature vector C i , the weight model parameters corresponding to the S units of virtual assets include unit virtual asset A i The corresponding weight model parameter D i , i is a positive integer less than or equal to S; the feature combination module 14 is used to combine the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets, and when the fused feature vectors corresponding to the S unit virtual assets are obtained, the feature combination module 14 is specifically used to perform the following operations:

[0194] Determine the identification requirement type and unit virtual asset A indicated by the business processing instruction i The correlation between virtual assets A based on the correlation i Assigning data association weights;

[0195] For the data feature vector C i and weight model parameter D i Splice and obtain data splicing features, and multiply the data splicing features and the data association weight to determine the unit virtual asset Ai The corresponding fused feature vector.

[0196] In a possible implementation, the target processing network is a pricing prediction network; the network determination module 15 is used to perform feature processing on the S fused feature vectors and the prompt text through the target processing network, and when the feature processing result is obtained, the network determination module 15 is specifically used to perform the following operations:

[0197] Extract features of the prompt text to obtain prompt text features;

[0198] Based on the search engine associated with the pricing prediction network, the environmental factor information of the object to be priced indicated by the prompt text is obtained;

[0199] Extract features of environmental factor information to obtain environmental factor features, fuse the environmental factor features, prompt text features and S fused feature vectors to obtain feature processing results;

[0200] The classification and recognition module 16 is used to classify and recognize the feature processing results. When a data recognition result for responding to the business processing instruction is obtained, the classification and recognition module 16 is specifically used to perform the following operations:

[0201] Obtain P price range labels corresponding to the pricing prediction network, and the estimated market information corresponding to the P price range labels, respectively, perform feature extraction on the P estimated market information, and obtain P price market features; P is a positive integer;

[0202] Perform cross-attention processing on the feature processing results and P price market features to obtain P first cross-attention scores;

[0203] The price range label corresponding to the estimated market information associated with the first target cross-attention score is determined as the data recognition result for responding to the business processing instruction; the first target cross-attention score is the largest cross-attention score among P first cross-attention scores.

[0204] In a possible implementation, the classification recognition module 16 is used to perform cross-attention processing on the feature processing result and the P price market features. When P first cross-attention scores are obtained, the classification recognition module 16 is specifically used to perform the following operations:

[0205] Combine P price market features to obtain price combination features;

[0206] The feature processing result is used as the query vector in the cross attention function, the price combination feature is used as the key vector in the cross attention function, and the price combination feature is used as the value vector in the cross attention function;

[0207] The product of the query vector and the key vector is determined as the first fused feature through the cross attention function;

[0208] The number of dimensions corresponding to the price combination feature is obtained, and the product of the first fusion feature and the inverse of the number of dimensions is determined as the second fusion feature;

[0209] Based on the activation sub-function in the cross attention function, the second fused feature is converted into the first activated feature;

[0210] The product of the first activation feature and the value vector determines the attention processing sequence; the attention processing sequence is composed of P first cross-attention scores.

[0211] In a possible implementation, the target processing network is a data analysis network; the network determination module 15 is used to perform feature processing on the S fused feature vectors and the prompt text through the target processing network. When the feature processing result is obtained, the network determination module 15 is specifically used to perform the following operations:

[0212] Extract features of the prompt text to obtain prompt text features;

[0213] Fuse the prompt text features and S fused feature vectors to obtain feature processing results;

[0214] The classification and recognition module 16 is used to classify and recognize the feature processing results. When a data recognition result for responding to the business processing instruction is obtained, the classification and recognition module 16 is specifically used to perform the following operations:

[0215] Obtain Q data classification labels corresponding to the data analysis network, and label association information corresponding to the Q data classification labels, respectively, perform feature extraction on the Q label association information, and obtain Q data classification features; Q is a positive integer;

[0216] Perform cross-attention processing on the feature processing results and Q data classification features to obtain Q second cross-attention scores;

[0217] The data classification label corresponding to the label association information associated with the second target cross-attention score is determined as the data recognition result for responding to the business processing instruction; the second target cross-attention score is the largest cross-attention score among the Q second cross-attention scores.

[0218] In a possible implementation, the data processing device 700 further includes a result determination module 17, and the result determination module 17 is specifically configured to perform the following operations:

[0219] Perform information retrieval on the data recognition results and prompt texts through a large language model to obtain information retrieval results;

[0220] Acquire a template library associated with the target processing network, and acquire a service template corresponding to the prompt text from the template library;

[0221] Determine the text to be updated in the business template, replace the text to be updated in the business template based on the data recognition result and the information retrieval result to obtain the updated business text, and determine the updated business text as the business result for the business processing instruction.

[0222] In a possible implementation, the result determination module 17 is further configured to perform the following operations:

[0223] If the identification requirement type indicated by the prompt text includes a drawing processing requirement, the drawing image type is identified according to the drawing processing requirement, an image template having the drawing image type is obtained from an image template library, and a Gaussian noise image is generated according to the image template and the initial noise data;

[0224] The data recognition result and the prompt text are input into the large language model, and the data recognition result and the prompt text are fused and expanded by the large language model to obtain the drawing instruction text;

[0225] The Gaussian noise image and the drawing instruction text are input into the Vincent graph model, the Gaussian noise image is feature extracted by the Vincent graph model to obtain Gaussian noise features, and the Gaussian noise features are forward diffused to obtain a forward noise vector;

[0226] The drawing instruction text is feature-encoded by the text-generated graph model to obtain the drawing text feature, and the Gaussian noise image is denoised according to the forward noise vector and the drawing text feature to obtain the drawing data graph;

[0227] The data identification result and the drawn data graph are determined as business results for the business processing instruction.

[0228] See also Figure 8 , Figure 8 A data processing device provided in an embodiment of the present application is shown in FIG. Figure 2 The data processing device 800 may be a computer program (including program code, etc.) running in a computer device. For example, the data processing device 800 may be an application software. The data processing device 800 may be used to execute the corresponding steps of the method provided in the embodiment of the present application. Figure 8 As shown, the data processing device 800 can be used to Figure 3 , Figure 4 and Figure 6 The computer device in the corresponding embodiment, specifically, the data processing device 800 may include: a sample acquisition module 21, a sample division module 22, a parameter allocation module 23, a sample feature combination module 24, a sample feature processing module 25, and a parameter adjustment module 26.

[0229] The sample acquisition module 21 is used to acquire sample processing instructions, virtual asset sample data for a sample aircraft and sample identification results, and input the sample processing instructions and the virtual asset sample data of the sample aircraft into the initial business processing model; the sample aircraft is composed of at least two sample components;

[0230] The sample partitioning module 22 is used to obtain the asset data classification rules, and based on the N business types and at least two sample components in the asset data classification rules, perform data partitioning on the virtual asset sample data of the sample aircraft to obtain R unit sample virtual assets, and determine the sample binary classification information corresponding to each unit sample virtual asset; the sample binary classification information is used to characterize the business type and sample component to which the unit sample virtual asset belongs; N and R are both positive integers;

[0231] A parameter allocation module 23 is used to allocate random initialization model parameters to the binary classification information corresponding to the R unit sample virtual assets, and determine the R initialization model parameters as an initialization model parameter set;

[0232] The sample feature combination module 24 is used to combine the sample feature vectors corresponding to the R unit sample virtual assets and the initialization model parameters corresponding to the R unit sample virtual assets to obtain the sample fusion feature vectors corresponding to the R unit sample virtual assets; a sample fusion feature vector is composed of a sample feature vector and an initialization model parameter;

[0233] The sample feature processing module 25 is used to generate a sample prompt text indicating the sample identification requirement type based on the sample processing instruction, determine a sample processing network corresponding to the sample identification requirement type from the initial business processing model based on the sample prompt text, perform feature processing on the R sample fusion feature vectors and the sample prompt text through the sample processing network to obtain a sample feature processing result, classify and identify the sample feature processing result, and obtain a sample prediction result for responding to the sample processing instruction;

[0234] The parameter adjustment module 26 is used to generate a model loss value based on the sample prediction results and the sample recognition results, and adjust each initialization model parameter in the initialization model parameter set and the parameters of the sample processing network based on the model loss value until the initial business processing model meets the model convergence conditions and obtains the target business processing model; the target business processing model includes a weight model parameter set, and the weight model parameter set includes the weight model parameters obtained after the initialization model parameters are adjusted.

[0235] See also Fig. 9 , Fig. 9Schematic diagram of the structure of a computer device provided in an embodiment of the present application. Fig. 9 As shown, the computer device 900 in the embodiment of the present application may include: a processor 901, a network interface 904 and a memory 905. In addition, the above-mentioned computer device 900 may also include: a user interface 903, and at least one communication bus 902. Among them, the communication bus 902 is used to realize the connection and communication between these components. Among them, the user interface 903 may include a display screen (Display), a keyboard (Keyboard), and the user interface 903 may optionally include a standard wired interface and a wireless interface. The network interface 904 may optionally include a standard wired interface and a wireless interface (such as a WI-FI interface). The memory 905 may be a high-speed RAM memory, or it may be a non-volatile memory (non-volatile memory), such as at least one disk storage. The memory 905 may optionally also be at least one storage device located away from the aforementioned processor 901. As Fig. 9 As shown, the memory 905 as a computer-readable storage medium may include an operating system, a network communication module, a user interface module, and a device control application program.

[0236] The network interface 904 may provide a network communication element; the user interface 903 is mainly used to provide an input interface for the user; and the processor 901 may be used to call the device control application stored in the memory 905 to perform the following operations:

[0237] Acquire the business processing instruction and the virtual asset associated data for the aircraft sent by the first business object, and input the business processing instruction and the virtual asset associated data for the aircraft into the target business processing model; the aircraft is composed of at least two components;

[0238] Obtain asset data classification rules, and divide the virtual asset associated data of the aircraft based on N business types and at least two components in the asset data classification rules to obtain S unit virtual assets, and determine the binary classification information corresponding to each unit virtual asset; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers;

[0239] Based on the binary classification information corresponding to the S unit virtual assets, respectively, the weight model parameters corresponding to the S unit virtual assets are obtained from the weight model parameter set in the target business processing model; the weight model parameter set includes H weight model parameters that respectively represent different impact weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer;

[0240] The data feature vectors corresponding to the S units of virtual assets and the weight model parameters corresponding to the S units of virtual assets are combined to obtain fused feature vectors corresponding to the S units of virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter;

[0241] Generate a prompt text for indicating the identification requirement type based on the business processing instruction, determine a target processing network corresponding to the identification requirement type from a target business processing model based on the prompt text, perform feature processing on the S fused feature vectors and the prompt text through the target processing network, and obtain a feature processing result;

[0242] The feature processing results are classified and identified to obtain data identification results for responding to business processing instructions.

[0243] In a possible implementation manner, the processor 901 is further configured to perform the following operations:

[0244] Obtaining a sample processing instruction, sample data of virtual assets for a sample aircraft, and a sample identification result, and inputting the sample processing instruction and the sample data of virtual assets for the sample aircraft into an initial business processing model; the sample aircraft is composed of at least two sample components;

[0245] Obtain asset data classification rules, and divide the virtual asset sample data of the sample aircraft based on N business types and at least two sample components in the asset data classification rules to obtain R unit sample virtual assets, and determine the sample binary classification information corresponding to each unit sample virtual asset; the sample binary classification information is used to characterize the business type and sample component to which the unit sample virtual asset belongs; N and R are both positive integers;

[0246] Assign random initialization model parameters to the binary classification information corresponding to the R unit sample virtual assets respectively, and determine the R initialization model parameters as the initialization model parameter set;

[0247] The sample feature vectors corresponding to the R unit sample virtual assets and the initialization model parameters corresponding to the R unit sample virtual assets are combined to obtain the sample fusion feature vectors corresponding to the R unit sample virtual assets; a sample fusion feature vector is composed of a sample feature vector and an initialization model parameter;

[0248] Generate a sample prompt text for indicating the sample identification requirement type based on the sample processing instruction, determine a sample processing network corresponding to the sample identification requirement type from the initial business processing model based on the sample prompt text, perform feature processing on R sample fusion feature vectors and the sample prompt text through the sample processing network to obtain a sample feature processing result, classify and identify the sample feature processing result, and obtain a sample prediction result for responding to the sample processing instruction;

[0249] Based on the sample prediction results and the sample recognition results, a model loss value is generated, and each initialization model parameter in the initialization model parameter set and the parameters of the sample processing network are adjusted based on the model loss value until the initial business processing model meets the model convergence conditions, thereby obtaining a target business processing model; the target business processing model includes a weight model parameter set, and the weight model parameter set includes weight model parameters obtained after adjusting the initialization model parameters.

[0250] In addition, it should be noted that: the embodiment of the present application also provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, the computer program is suitable for being loaded and executed by the processor Figure 3 , Figure 4 or Figure 6 For details on the methods provided in each step, please refer to the Figure 3 , Figure 4 or Figure 6 The implementation methods provided in each step are not repeated here. In addition, the description of the beneficial effects of adopting the same method is not repeated. For technical details not disclosed in the computer-readable storage medium embodiment involved in this application, please refer to the description of the method embodiment of this application. As an example, the computer program can be deployed to execute on one computer device, or on multiple computer devices located in one place, or on multiple computer devices distributed in multiple locations and interconnected by a communication network.

[0251] The computer-readable storage medium may be the device provided in any of the aforementioned embodiments or the internal storage unit of the computer device, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (smart media card, SMC), a secure digital (secure digital, SD) card, a flash card (flash card), etc. equipped on the computer device. Further, the computer-readable storage medium may also include both the internal storage unit of the computer device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the computer device. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0252] The present application also provides a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes Figure 3 , Figure 4 or Figure 6 The methods provided in the various optional methods are therefore not described in detail here.

[0253] The terms "first", "second", etc. in the description, claims, and drawings of the embodiments of the present application are used to distinguish different objects, rather than to describe a specific order. In addition, the term "comprising" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, device, product, or equipment that includes a series of steps or units is not limited to the listed steps or modules, but optionally includes steps or modules that are not listed, or optionally includes other step units inherent to these processes, methods, devices, products, or equipment.

[0254] In the embodiments of the present application, the term "module" refers to a computer program or a part of a computer program with a predetermined function, and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0255] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in this description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0256] The method and related apparatus provided by the embodiment of the present application are described with reference to the method flow chart and / or structural diagram provided by the embodiment of the present application. Specifically, each process and / or box in the method flow chart and / or structural diagram, as well as the combination of the processes and / or boxes in the flow chart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable device to generate a machine, so that the instructions executed by the processor of the computer or other programmable device generate instructions for implementing the process in the process. Figure 1 A process or multiple processes and / or structures Figure 1 The computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including the instruction device, or are transmitted through a computer-readable storage medium. Computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The instruction device is implemented in the process Figure 1 A process or multiple processes and / or structures Figure 1 These computer program instructions can also be loaded onto a computer or other programmable device so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide the functions for implementing the process. Figure 1 A flow or multiple flows and / or structures illustrate the steps of the functions specified in one block or multiple blocks.

[0257] The steps in the method of the embodiment of the present application can be adjusted in order, combined and deleted according to actual needs.

[0258] The modules in the device of the embodiment of the present application can be merged, divided and deleted according to actual needs.

[0259] The above disclosure is only the preferred embodiment of the present application, which certainly cannot be used to limit the scope of rights of the present application. Therefore, equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A data processing method, characterized in that: The method comprises: Acquire a business processing instruction and virtual asset associated data for an aircraft sent by a first business object, and input the business processing instruction and the virtual asset associated data for the aircraft into a target business processing model; the aircraft is composed of at least two components; Acquire asset data classification rules, and divide the virtual asset associated data of the aircraft based on the N business types and the at least two components in the asset data classification rules to obtain S unit virtual assets, and determine the binary classification information corresponding to each unit virtual asset; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers; Based on the binary classification information corresponding to the S unit virtual assets, respectively, obtaining the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model; the weight model parameter set includes H weight model parameters that respectively represent different impact weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer; Combining the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets to obtain fused feature vectors corresponding to the S unit virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter; Generate a prompt text for indicating the identification requirement type based on the business processing instruction, determine a target processing network corresponding to the identification requirement type from the target business processing model based on the prompt text, and perform feature processing on S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result; The feature processing results are classified and identified to obtain data identification results for responding to the business processing instructions.

2. The method according to claim 1, characterized in that The step of dividing the virtual asset associated data of the aircraft based on the N types of business in the asset data classification rule and the at least two components to obtain S unit virtual assets and determining the binary classification information corresponding to each unit virtual asset includes: The virtual asset associated data of the aircraft is classified according to the N business types in the asset data classification rule to obtain M types of classification data; the business types corresponding to the M types of classification data belong to the N business types, and M is a positive integer less than or equal to N; In each type of classified data, each type of classified data is divided into components based on the at least two components to obtain S unit virtual assets; one unit virtual asset corresponds to one business type and one component; The business types and components corresponding to the S unit virtual assets are respectively obtained, and the business types and components corresponding to the S unit virtual assets are combined to obtain binary classification information corresponding to the S unit virtual assets respectively.

3. The method according to claim 2, characterized in that The N types of business types include asset business data types, basic business data types, and transaction business data types; if M is equal to N, the M types of classified data include first classified data, second classified data, and third classified data; The virtual asset associated data of the aircraft is classified according to the N business types in the asset data classification rule to obtain M types of classification data, including: Acquire the data key fields corresponding to the asset business data type; the data key fields include the usage record key field, the maintenance record key field and the depreciation record key field; In the virtual asset associated data of the aircraft, historical usage records corresponding to the usage record key field, historical maintenance records corresponding to the maintenance record key field, and historical damage records corresponding to the damage record key field are obtained, and the historical usage records, the historical maintenance records, and the historical damage records are determined as first classification data corresponding to the asset business data type; Acquire a factory key field corresponding to the basic business data type, acquire factory-related data corresponding to the factory key field in the virtual asset-related data of the aircraft, and determine the factory-related data as the second classification data corresponding to the basic business data type; Obtain a transaction key field corresponding to the transaction business data type, obtain transaction-related data corresponding to the transaction key field in the virtual asset-related data of the aircraft, and determine the transaction-related data as third classification data corresponding to the transaction business data type.

4. The method according to claim 1, characterized in that: The binary classification information corresponding to the S units of virtual assets includes binary classification information B i , i is a positive integer less than or equal to S; the step of obtaining the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model based on the binary classification information corresponding to the S unit virtual assets respectively includes: Acquire a weight model parameter set and a parameter matching rule in the target business processing model; the parameter matching rule is used to indicate a mapping relationship between a weight model parameter in the weight model parameter set and binary classification information; Determining a binary classification information set based on the weight model parameter set and the parameter matching rule; The binary classification information set is combined with the binary classification information B i The matched bigram classification information is determined as the target bigram classification information; Based on the target binary classification information and the parameter matching rule, the target weight model parameter corresponding to the target binary classification information is obtained from the weight model parameter, and the target weight model parameter is determined as the binary classification information B i The corresponding weight model parameters.

5. The method according to claim 1, characterized in that The S units of virtual assets include unit virtual asset A i The data feature vectors corresponding to the S unit virtual assets include unit virtual asset A i The corresponding data feature vector C i The weight model parameters corresponding to the S unit virtual assets include unit virtual asset A i The corresponding weight model parameter D i , i is a positive integer less than or equal to S; the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets are combined to obtain the fused feature vectors corresponding to the S unit virtual assets, including: Determine the identification requirement type indicated by the business processing instruction and the unit virtual asset A i The correlation between the unit virtual asset A and the unit virtual asset A is based on the correlation i Assigning data association weights; For the data feature vector C i and the weight model parameter D i Splicing is performed to obtain data splicing features, and the product of the data splicing features and the data association weight is determined as the unit virtual asset A i The corresponding fused feature vector.

6. The method according to claim 1, characterized in that The target processing network is a pricing prediction network; the target processing network performs feature processing on the S fused feature vectors and the prompt text to obtain a feature processing result, including: Extracting features from the prompt text to obtain prompt text features; Based on a search engine associated with the pricing prediction network, obtaining environmental factor information of the object to be priced indicated by the prompt text; Extracting features from the environmental factor information to obtain environmental factor features, and fusing the environmental factor features, the prompt text features, and the S fused feature vectors to obtain a feature processing result; The classifying and identifying the feature processing results to obtain a data identification result for responding to the business processing instruction includes: Obtain P price range labels corresponding to the pricing prediction network, and estimated market information corresponding to the P price range labels, respectively, and perform feature extraction on the P estimated market information to obtain P price market features; P is a positive integer; Performing cross-attention processing on the feature processing result and the P price market features to obtain P first cross-attention scores; The price range label corresponding to the estimated market information associated with the first target cross-attention score is determined as the data recognition result for responding to the business processing instruction; the first target cross-attention score is the largest first cross-attention score among the P first cross-attention scores.

7. The method according to claim 6, characterized in that The cross-attention processing is performed on the feature processing result and the P price market features to obtain P first cross-attention scores, including: Combining the P price market characteristics to obtain a price combination characteristic; Using the feature processing result as a query vector in a cross-attention function, using the price combination feature as a key vector in the cross-attention function, and using the price combination feature as a value vector in the cross-attention function; Determining a product of the query vector and the key vector as a first fused feature through the cross attention function; Acquire the number of dimensions corresponding to the price combination feature, and determine the product of the first fusion feature and the reciprocal of the number of dimensions as the second fusion feature; Based on the activation sub-function in the cross attention function, converting the second fused feature into a first activation feature; The product of the first activation feature and the value vector is used to determine an attention processing sequence; the attention processing sequence is composed of P first cross-attention scores.

8. The method according to claim 1, characterized in that The target processing network is a data analysis network; the feature processing is performed on the S fused feature vectors and the prompt text by the target processing network to obtain a feature processing result, including: Extracting features from the prompt text to obtain prompt text features; Fusing the prompt text feature and the S fused feature vectors to obtain a feature processing result; The classifying and identifying the feature processing results to obtain a data identification result for responding to the business processing instruction includes: Obtain Q data classification labels corresponding to the data analysis network, and label association information corresponding to the Q data classification labels, respectively, and perform feature extraction on the Q label association information to obtain Q data classification features; Q is a positive integer; Performing cross attention processing on the feature processing result and the Q data classification features to obtain Q second cross attention scores; The data classification label corresponding to the label association information associated with the second target cross-attention score is determined as the data recognition result for responding to the business processing instruction; the second target cross-attention score is the largest second cross-attention score among the Q second cross-attention scores.

9. The method according to claim 1, characterized in that: The method further comprises: Performing information retrieval on the data recognition result and the prompt text through a large language model to obtain an information retrieval result; Acquire a template library associated with the target processing network, and acquire a service template corresponding to the prompt text from the template library; Determine the text to be updated in the business template, replace the text to be updated in the business template based on the data recognition result and the information retrieval result to obtain updated business text, and determine the updated business text as the business result for the business processing instruction.

10. The method according to claim 1, characterized in that The method further comprises: If the identification requirement type indicated by the prompt text includes a drawing processing requirement, identifying a drawing image type according to the drawing processing requirement, acquiring an image template having the drawing image type from an image template library, and generating a Gaussian noise image according to the image template and initial noise data; Inputting the data recognition result and the prompt text into a large language model, fusing and expanding the data recognition result and the prompt text through the large language model to obtain a drawing instruction text; Inputting the Gaussian noise image and the drawing instruction text into a Vincent graph model, performing feature extraction on the Gaussian noise image through the Vincent graph model to obtain Gaussian noise features, and performing forward diffusion processing on the Gaussian noise features to obtain a forward noise vector; Performing feature encoding on the drawing instruction text by using the text graph model to obtain drawing text features, and performing denoising processing on the Gaussian noise image according to the forward noise vector and the drawing text features to obtain a drawing data graph; The data recognition result and the drawn data graph are determined as business results for the business processing instruction.

11. A data processing method, characterized in that: The method comprises: Obtaining a sample processing instruction, sample data of virtual assets for a sample aircraft, and a sample identification result, and inputting the sample processing instruction and the sample data of virtual assets for the sample aircraft into an initial business processing model; the sample aircraft is composed of at least two sample components; Acquire asset data classification rules, and divide the virtual asset sample data of the sample aircraft based on the N business types and the at least two sample components in the asset data classification rules to obtain R unit sample virtual assets, and determine sample binary classification information corresponding to each unit sample virtual asset; the sample binary classification information is used to characterize the business type and sample component to which the unit sample virtual asset belongs; N and R are both positive integers; Assigning random initialization model parameters to the binary classification information corresponding to the R unit sample virtual assets respectively, and determining the R initialization model parameters as an initialization model parameter set; The sample feature vectors corresponding to the R unit sample virtual assets and the initialization model parameters corresponding to the R unit sample virtual assets are combined to obtain sample fusion feature vectors corresponding to the R unit sample virtual assets; a sample fusion feature vector is composed of a sample feature vector and an initialization model parameter; Generate a sample prompt text for indicating a sample identification requirement type based on the sample processing instruction, determine a sample processing network corresponding to the sample identification requirement type from the initial business processing model based on the sample prompt text, perform feature processing on R sample fusion feature vectors and the sample prompt text through the sample processing network to obtain a sample feature processing result, classify and identify the sample feature processing result, and obtain a sample prediction result for responding to the sample processing instruction; Based on the sample prediction result and the sample identification result, a model loss value is generated, and based on the model loss value, each initialization model parameter in the initialization model parameter set and the parameters of the sample processing network are adjusted until the initial business processing model satisfies the model convergence condition, thereby obtaining a target business processing model; the target business processing model includes a weight model parameter set, and the weight model parameter set includes the weight model parameters obtained after the initialization model parameters are adjusted.

12. A data processing device, characterized in that: The device comprises: A data acquisition module, used to acquire a business processing instruction sent by the first business object and virtual asset associated data for the aircraft, and input the business processing instruction and the virtual asset associated data of the aircraft into a target business processing model; the aircraft is composed of at least two components; A data partitioning module is used to obtain an asset data classification rule, and based on the N types of business types and the at least two components in the asset data classification rule, perform data partitioning on the virtual asset associated data of the aircraft to obtain S unit virtual assets, and determine the binary classification information corresponding to each unit virtual asset; the binary classification information is used to characterize the business type and component to which the unit virtual asset belongs; N and S are both positive integers; A parameter acquisition module, configured to acquire, based on the binary classification information corresponding to the S unit virtual assets, the weight model parameters corresponding to the S unit virtual assets from the weight model parameter set in the target business processing model; the weight model parameter set includes H weight model parameters representing different influence weights, each weight model parameter corresponds to a binary classification information, and H is a positive integer; A feature combination module, used to combine the data feature vectors corresponding to the S unit virtual assets and the weight model parameters corresponding to the S unit virtual assets to obtain fused feature vectors corresponding to the S unit virtual assets; a fused feature vector is composed of a data feature vector and a weight model parameter; a network determination module, configured to generate a prompt text indicating the identification requirement type based on the business processing instruction, determine a target processing network corresponding to the identification requirement type from the target business processing model based on the prompt text, and perform feature processing on the S fused feature vectors and the prompt text through the target processing network to obtain a feature processing result; The classification and identification module is used to classify and identify the feature processing results to obtain data identification results for responding to the business processing instructions.

13. A data processing device, characterized in that: The device comprises: A sample acquisition module, used to acquire sample processing instructions, virtual asset sample data for a sample aircraft and a sample recognition result, and input the sample processing instructions and the virtual asset sample data of the sample aircraft into an initial business processing model; the sample aircraft is composed of at least two sample components; The sample division module is used to obtain the asset data classification rules, and based on the N business types and the at least two sample components in the asset data classification rules, divide the virtual asset sample data of the sample aircraft to obtain R unit sample virtual assets, and determine the sample binary classification information corresponding to each unit sample virtual asset; the sample binary classification information is used to characterize the business type and sample component to which the unit sample virtual asset belongs; N and R are both positive integers; A parameter allocation module, used to allocate random initialization model parameters to the binary classification information corresponding to the R unit sample virtual assets, and determine the R initialization model parameters as an initialization model parameter set; A sample feature combination module is used to combine the sample feature vectors corresponding to the R unit sample virtual assets and the initialization model parameters corresponding to the R unit sample virtual assets to obtain sample fusion feature vectors corresponding to the R unit sample virtual assets; a sample fusion feature vector is composed of a sample feature vector and an initialization model parameter; A sample feature processing module, used to generate a sample prompt text indicating a sample identification requirement type based on the sample processing instruction, determine a sample processing network corresponding to the sample identification requirement type from the initial business processing model based on the sample prompt text, perform feature processing on R sample fusion feature vectors and the sample prompt text through the sample processing network to obtain a sample feature processing result, classify and identify the sample feature processing result, and obtain a sample prediction result for responding to the sample processing instruction; A parameter adjustment module is used to generate a model loss value based on the sample prediction result and the sample identification result, and to adjust each initialization model parameter in the initialization model parameter set and the parameters of the sample processing network based on the model loss value until the initial business processing model satisfies the model convergence condition and obtains a target business processing model; the target business processing model includes a weight model parameter set, and the weight model parameter set includes the weight model parameters obtained after the initialization model parameters are adjusted.

14. A computer device, characterized in that: Includes processor, memory, input and output interfaces; The processor is connected to the memory and the input / output interface respectively, wherein the input / output interface is used to receive and output data, the memory is used to store a computer program, and the processor is used to call the computer program so that the computer device executes the method described in any one of claims 1-11.

15. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded and executed by a processor, so that a computer device having the processor executes the method according to any one of claims 1 to 11.

16. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method according to any one of claims 1 to 11 is implemented.

Images