A data resource allocation method, device, equipment, storage medium and product

Abstract

The application discloses a data resource allocation method, comprising: acquiring multi-modal user data and uploading to a trusted data space; the multi-modal user data comprises user behavior logs, user attribute data, historical business response records and user geographic location trajectories; feature extraction is performed on the user data from multiple dimensions, and weighted average is performed on the characteristic values of the dimensions to obtain an initial value of the user data; the characteristic values of the dimensions and the initial value are input into a data value prediction model, and the potential commercial value of the user data and the business scene matching degree are output; a bipartite graph of the user data and the business scene is established, matching of the data and the business scene is performed in the bipartite graph based on the potential commercial value and the business scene matching degree, and an optimal matching data set corresponding to each business scene is output. The application can comprehensively and accurately evaluate the value of the user data, improve the adaptability of the data and the business scene, and make the data resources more reasonably and efficiently allocated.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a data resource allocation method, apparatus, device, storage medium, and product. Background Technology

[0002] Currently, business intelligence systems are widely used in market data value analysis and business trend prediction, playing an important role in industries such as retail, e-commerce, finance, and telecommunications operators. Most existing business intelligence systems collect massive amounts of historical data, use rule engines or basic machine learning algorithms to process the data, and then mine user preferences, predict business trends, and guide business operations, thus achieving the initial mining and utilization of data value.

[0003] However, in practical applications, existing solutions have a single dimension for data value assessment, using only simple rules or single-level models for calculation. This makes it difficult to fully quantify the potential value of data and accurately match diverse business scenarios. It is prone to data mismatch and resource waste, making it difficult to fully release the value of data and resulting in low business conversion efficiency. Summary of the Invention

[0004] This invention provides a data resource allocation method that can comprehensively and accurately assess the value of user data, improve the adaptability of data to business scenarios, and enable data resources to be allocated more rationally and efficiently.

[0005] In a first aspect, embodiments of the present invention provide a data resource allocation method, including: Acquire multimodal user data and upload it to a trusted data space; the multimodal user data includes, but is not limited to, user behavior logs, user attribute data, historical business response records, and user geographic location trajectories; The user data is feature extracted from multiple dimensions, and the initial value of the user data is obtained by weighted averaging of the feature values ​​of each dimension. The feature values ​​of each dimension and the initial value are input into a pre-built data value prediction model, which outputs the potential commercial value of user data and the business scenario matching degree for each business scenario. Establish a bipartite graph of user data and business scenarios. Based on the potential business value and the matching degree of the business scenarios, perform data matching with business scenarios in the bipartite graph, and output the optimal matching dataset corresponding to each business scenario.

[0006] Furthermore, the construction process of the data value prediction model includes: Obtain the multi-dimensional features and real value tags of historical user data; wherein, the real value tags include potential business value tags and business scenario matching tags; A regression analysis model was used to fit the linear relationship between each dimension feature and the real value label to obtain preliminary value fitting results. A decision tree model is used to learn nonlinear rules based on the features of each dimension and the preliminary value fitting results to obtain a single decision tree evaluation model. The random forest algorithm is used to integrate multiple decision tree models with different structures to obtain a data value prediction model.

[0007] Furthermore, the step of establishing a bipartite graph of user data and business scenarios involves matching data with business scenarios within the bipartite graph based on the potential business value and the matching degree of the business scenarios, and outputting the optimal matching dataset corresponding to each business scenario, including: By using user data as data nodes and business scenarios as scenario nodes, a bipartite graph structure is constructed. A bipartite graph matching algorithm based on an attention mechanism is used to calculate the total matching value between each data node and each scene node; Using the maximization of the total matching value as the matching optimization objective, the optimal matching scheme of the bipartite graph is solved; Based on the optimal matching scheme, output the optimal matching data list corresponding to each business scenario.

[0008] Furthermore, the total matching value is the product of the potential business value and the matching degree of the business scenario.

[0009] Furthermore, the method also includes: During the process of uploading the multimodal user data to the trusted data space, an asymmetric encryption algorithm is used to encrypt and protect the data, and digital signature technology is used to verify the authenticity of the data source and the integrity of the data. The metadata of the user data, including data type, generation time, data source, signature digest, and other information, is written into a blockchain smart contract for storage.

[0010] Furthermore, the method also includes: Based on the overall revenue actually generated after matching data with business scenarios, and using the Shapley value and kernel function in cooperative game theory, the profit distribution share of each participating entity is calculated according to the marginal contribution of each participating entity to the overall revenue. The profit distribution share will be dynamically adjusted based on changes in the market environment, data circulation environment, and cooperative relationships among participating entities.

[0011] In a second aspect, embodiments of the present invention provide a data resource allocation apparatus, comprising: The data acquisition module is used to acquire multimodal user data and upload it to a trusted data space; the multimodal user data includes, but is not limited to, user behavior logs, user attribute data, historical business response records, and user geographic location trajectories. The value prediction module is used to extract features from the user data from multiple dimensions, input the extracted features into a pre-built data value prediction model, and output the potential commercial value of the user data and the business scenario matching degree for each business scenario. The data matching module is used to establish a bipartite graph of user data and business scenarios. Based on the potential business value and the matching degree of the business scenarios, it performs data matching with business scenarios in the bipartite graph and outputs the optimal matching dataset corresponding to each business scenario.

[0012] Thirdly, embodiments of the present invention provide an electronic device, comprising: Memory, used to store computer programs; A processor for executing the computer program; Wherein, when the processor executes the computer program, it implements the data resource allocation method described in any of the first aspects above.

[0013] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing a computer program, which, when executed, implements the data resource allocation method described in any of the first aspects above.

[0014] Fifthly, embodiments of the present invention provide a computer program product, including computer instructions, which, when executed by a processor, implement the data resource allocation method described in any of the first aspects above.

[0015] Compared with existing technologies, the data resource allocation method provided in this invention has the following advantages: It acquires multimodal user data and uploads it to a trusted data space; the multimodal user data includes, but is not limited to, user behavior logs, user attribute data, historical business response records, and user geographic location trajectories; it extracts features from the user data from multiple dimensions and performs a weighted average of the feature values ​​of each dimension to obtain the initial value of the user data; it inputs the feature values ​​of each dimension and the initial value into a pre-constructed data value prediction model, outputting the potential commercial value of the user data and the business scenario matching degree for each business scenario; it establishes a bipartite graph of user data and business scenarios, performs data-business scenario matching based on the potential commercial value and the business scenario matching degree in the bipartite graph, and outputs the optimal matching dataset corresponding to each business scenario. This invention can comprehensively and accurately evaluate the value of user data, improve the adaptability of data to business scenarios, and enable more reasonable and efficient allocation of data resources. Attached Figure Description

[0016] To more clearly illustrate the technical features of the embodiments of the present invention, the drawings used in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a data resource allocation method provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a data resource allocation device provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0021] In a first aspect, embodiments of the present invention provide a data resource allocation method, see [link to relevant documentation]. Figure 1 This is a flowchart illustrating an embodiment of a data resource allocation method provided by the present invention.

[0022] like Figure 1 As shown, the method includes the following steps: S1: Acquire multimodal user data and upload it to a trusted data space; the multimodal user data includes, but is not limited to, user behavior logs, user attribute data, historical business response records, and user geographic location trajectories; Specifically, multimodal user data is collected, including but not limited to: user behavior logs, user attribute data, historical business response records, and user geographic location trajectories. User behavior logs refer to various behavioral records generated by users on business platforms and related apps, including operation frequency, dwell time, click records, browsing records, search content, etc., used to reflect user activity and behavioral preferences; user attribute data refers to the user's basic identity and business-related attributes, including age, gender, online time, package type, consumption level, business transaction history, etc., used to create a basic user profile; historical business response records refer to the user's past feedback results to marketing pushes, business activities, and service notifications, including records of processing, rejection, ignoring, and inquiries, used to measure user conversion potential; user geographic location trajectories refer to location information within the user's authorized range, including permanent residence, activity range, travel trajectory, etc., used to adapt to regionally relevant business scenarios.

[0023] After completing the collection of multimodal user data, the data is initially cleaned to remove duplicate, missing, and abnormal invalid entries. The cleaned and complete data is then uploaded to the Trusted Data Space, a compliant storage environment with secure encryption, access control, and traceability capabilities. It enables isolated storage and unified management of data, preventing data leakage, tampering, and unauthorized access.

[0024] S2: Extract features from the user data from multiple dimensions, and calculate the initial value of the user data by weighted averaging of the feature values ​​of each dimension; Specifically, starting from the operator's business scenarios, user data features are extracted from four core dimensions: data frequency, data activity, data quality, and user behavior. Data frequency refers to the frequency at which user data is collected and recorded per unit of time. Generally, the higher the data collection frequency, the more completely it reflects the changing patterns of user behavior, and the higher its potential value. Data activity refers to the level of user activity corresponding to the data, or the dynamic update level of the data itself. The higher the activity level of the data, the stronger its supporting value for real-time business decisions and precision marketing. Data quality refers to the usability of the data, mainly including three indicators: data completeness, accuracy, and consistency. High-quality data can provide reliable decision support for business operations, while low-quality data will reduce the effectiveness of analysis and matching. User behavior refers to the interactive behaviors of users in response to business. The value of behavior is measured through key behavioral indicators, directly reflecting the user's willingness to convert business and the level of data value.

[0025] For the four data dimensions, quantification calculations were performed using corresponding formulas to obtain individual value scores for each dimension. All calculations were conducted within standardized numerical ranges to ensure the comparability and fusion of scores across dimensions. The data frequency value was calculated using the following formula: ; in, This represents the frequency value of data, specifically the average frequency of events occurring within a unit of time window. This represents the frequency of data events occurring within the i-th time window, and n represents the total number of time windows for statistics.

[0026] Data activity value is calculated using the following formula: ; in, This represents the value of data activity, specifically the average number of active users per time window. This represents the number of active users within the i-th time window, and n represents the total number of time windows.

[0027] Data quality value is calculated using a weighted summation of multiple indicators, as shown in the following formula: ; in, Indicates the value of data quality. This indicates data integrity, specifically the proportion of valid data to the total data. This indicates data accuracy, specifically the proportion of data that is free of errors and anomalies. This indicates data consistency, specifically the degree to which data formats and definitions are consistent. These are the weighting coefficients for the three indicators, with the sum of the weights being 1, and are pre-set based on business experience and actual needs.

[0028] User behavior value is calculated using a weighted average of core interaction metrics, as shown in the following formula: ; in, Indicates the value of user behavior. This refers to the click-through rate, which is the proportion of times a user clicks on business content out of the number of times it is displayed. This indicates user engagement time, that is, the total time users spend browsing and interacting. The corresponding weighting coefficients can be flexibly set according to the business focus.

[0029] After quantifying the value across the four dimensions, a weighted average method is used to combine the value scores of each dimension to obtain the initial potential value of a single piece of user data. The calculation formula is shown below: ; in, , The weighting coefficients are for four dimensions: data frequency, data activity, data quality, and user behavior. The weights are pre-set based on the degree of impact of each dimension on business value.

[0030] S3: Input the feature values ​​of each dimension and the initial value into the pre-built data value prediction model, and output the potential commercial value of user data and the business scenario matching degree for each business scenario; Specifically, based on the multi-dimensional initial value assessment, a pre-trained data value prediction model is introduced. This model uses massive amounts of historical user data as training samples and is an ensemble learning model obtained through feature standardization, preliminary fitting via regression analysis, non-linear learning via decision trees, and ensemble optimization via random forests. It can uncover complex relationships between features, overcome the one-sidedness of manual weighted evaluation, adapt to various complex data scenarios, and output more accurate results that fit actual business needs. The output results include the potential commercial value of user data and the business scenario matching degree for each business scenario. Potential commercial value represents the comprehensive commercial value that the user data can generate after being applied to business applications, such as profit and conversion effects. The higher the score, the greater the commercial potential of the data. Business scenario matching degree represents the degree of fit between the user data features and the target user profile and business needs of the corresponding scenario.

[0031] S4: Establish a bipartite graph of user data and business scenarios. Based on the potential business value and the matching degree of the business scenarios, perform data matching with business scenarios in the bipartite graph, and output the optimal matching dataset corresponding to each business scenario.

[0032] Specifically, a bipartite graph structure corresponding to user data and business scenarios is constructed. Individual user data points that have undergone value assessment are set as data nodes, and various operator business scenarios are set as scenario nodes. The potential commercial value of user data and the matching degree of business scenarios are used as core indicators. The optimal matching algorithm is executed in the constructed bipartite graph. The matching process takes maximizing the global total value as the optimization goal, comprehensively considering the commercial potential of the data itself and the scenario adaptability. Finally, the optimal matching dataset corresponding to each business scenario is output. The user data in this dataset are all highly adapted to the corresponding scenario and have high commercial potential. They can be directly used for subsequent business implementation, precision marketing, and other links, which can significantly improve business conversion efficiency and maximize the value of data resources.

[0033] In summary, this invention optimizes the entire process of multimodal user data acquisition, from raw collection to scenario-based intelligent matching. By collecting multimodal user data and uploading it to a trusted data space, it ensures the secure and compliant flow of user privacy data. By combining weighted feature extraction and prediction models to quantify data value, it solves the problems of strong subjectivity and single dimension in traditional data value assessment, effectively improving the accuracy and objectivity of value assessment. The bipartite graph optimization matching algorithm based on potential commercial value and business scenario matching degree can accurately identify the optimal data combination in different business scenarios, significantly improving the allocation efficiency and commercial conversion effect of data resources, and maximizing the value of data resources.

[0034] In one optional implementation, the process of constructing the data value prediction model includes: Obtain the multi-dimensional features and real value tags of historical user data; wherein, the real value tags include potential business value tags and business scenario matching tags; A regression analysis model was used to fit the linear relationship between each dimension feature and the real value label to obtain preliminary value fitting results. A decision tree model is used to learn nonlinear rules based on the features of each dimension and the preliminary value fitting results to obtain a single decision tree evaluation model. The random forest algorithm is used to integrate multiple decision tree models with different structures to obtain a data value prediction model.

[0035] Specifically, historical user data is first acquired, including user behavior logs, user attribute data, historical business response records, user geographic location trajectories, and other information. The historical user data is then standardized to eliminate differences in the units and magnitudes of different features. After preprocessing, multi-dimensional features are extracted as input features for the machine learning model. At the same time, each piece of historical data is labeled with a real value tag as a supervision target for model training, including potential business value tags and business scenario matching tags.

[0036] A regression analysis model is used, preferably linear regression or multinomial regression, to train the standardized multidimensional features and real value labels, capture the basic trend between features and value, fit the linear relationship between features and target variables, and obtain preliminary value fitting results.

[0037] Using a decision tree model, the multi-dimensional features and preliminary value fitting results obtained in the previous step are used as input for nonlinear rule learning. The decision tree, through feature splitting conditions, explores the complex relationships generated by the superposition and intersection of multiple features, and generates a highly interpretable single decision tree evaluation model.

[0038] By employing the random forest algorithm, multiple decision tree models with different structures and training samples are integrated. Through multi-tree voting and comprehensive decision-making, the prediction results are optimized. Random forest solves the problems of overfitting and weak generalization ability of single decision trees. It is especially suitable for scenarios with complex nonlinear relationships between features, greatly improving the stability and prediction accuracy of the model, and finally obtaining a data value prediction model.

[0039] This embodiment employs a progressive model building approach using regression analysis, decision trees, and random forests. This approach can quickly capture the basic linear trend between features and value, while also deeply exploring the complex nonlinear relationships resulting from the cross-over of multiple features. It balances model prediction accuracy and interpretability, significantly improving the stability and accuracy of value assessment. Ultimately, it can accurately output the potential commercial value and scenario matching of data, providing reliable support for the accurate matching of data with business scenarios in the future.

[0040] In one optional implementation, the step of establishing a bipartite graph of user data and business scenarios, performing data-business scenario matching in the bipartite graph based on the potential business value and the matching degree of the business scenarios, and outputting the optimal matching dataset corresponding to each business scenario, includes: By using user data as data nodes and business scenarios as scenario nodes, a bipartite graph structure is constructed. A bipartite graph matching algorithm based on an attention mechanism is used to calculate the total matching value between each data node and each scene node; Using the maximization of the total matching value as the matching optimization objective, the optimal matching scheme of the bipartite graph is solved; Based on the optimal matching scheme, output the optimal matching data list corresponding to each business scenario.

[0041] Specifically, each individual user data point that has undergone value assessment is used as a data node, and Transformer encoding is performed on the user data to obtain a data embedding vector. , recorded as ,in This represents the embedding vector corresponding to the i-th user data, where n is the total number of user data entries. Various operator business scenarios (such as 5G package upgrades, home broadband promotion, and data package marketing) are used as scenario nodes. The scenario embedding vector is trained based on business scenario labels and historical business data. , recorded as ,in Let m represent the embedding vector corresponding to the j-th business scenario, where m is the total number of business scenarios.

[0042] Establish connecting edges between data nodes and scene nodes to form an undirected bipartite graph. Calculate the total matching value between data nodes and scene nodes. With maximizing the total matching value as the core optimization objective, use an attention-aware bipartite graph matching algorithm to solve for the optimal matching in the bipartite graph, and select the matching combination with the highest global value. After the solution is completed, organize and output the optimal matching data list corresponding to each business scenario based on the optimal matching scheme to complete the accurate allocation of data resources.

[0043] This embodiment constructs a bipartite graph structure and employs an attention-aware bipartite graph matching algorithm to combine data value with scenario adaptability. The allocation is completed with the goal of maximizing the total matching value. This not only achieves accurate matching between user data and corresponding business scenarios, avoiding data mismatch and waste, but also fully leverages the conversion potential of high-value data to improve overall business revenue. At the same time, the algorithm is adaptable to the allocation needs of multiple scenarios and large-scale data, balancing matching accuracy and execution efficiency, and realizing the rational and efficient allocation of data resources.

[0044] In one alternative implementation, the total matching value is the product of the potential business value and the business scenario matching degree.

[0045] Specifically, the formula for the optimal matching target is as follows: ; Where M represents the set of matching pairs between data and business scenarios, and each element in the set represents the matching relationship between a piece of user data and a business scenario. This represents the matching action of assigning the i-th user's data to the j-th business scenario. This represents the matching degree between the i-th data node and the j-th scene node. This represents the potential commercial value of the i-th user's data.

[0046] In an optional implementation, the method further includes: During the process of uploading the multimodal user data to the trusted data space, an asymmetric encryption algorithm is used to encrypt and protect the data, and digital signature technology is used to verify the authenticity of the data source and the integrity of the data. The metadata of the user data, including data type, generation time, data source, signature digest, and other information, is written into a blockchain smart contract for storage.

[0047] Specifically, to ensure the security of multimodal user data throughout the entire process of uploading, storing, and accessing, this embodiment combines asymmetric encryption, digital signatures, and blockchain technology to build a complete data security protection system. This system enables trusted data storage, source verification, and refined access control, implementing security protection throughout the entire process of uploading data to a trusted data space and preventing data leakage, tampering, and unauthorized access.

[0048] First, the data is protected by an asymmetric encryption algorithm, such as RSA or ECC elliptic curve encryption. Asymmetric encryption uses a public-private key pairing method. Each data provider has a unique and corresponding private key and public key. The private key is kept by the data provider and is not disclosed to the public, while the public key can be made available to authorized users.

[0049] During the data upload phase, the data provider uses its private key to encrypt the multimodal user data, generating encrypted ciphertext. When the encrypted data is transmitted to the trusted data space, the receiver uses the corresponding public key to decrypt it and verify whether the data has been tampered with during transmission. After the data is stored in the trusted data space, it is stored in encrypted form throughout the entire process. Only authorized entities with decryption privileges can call the corresponding key to decrypt and read the data content. Even if the data is illegally intercepted during the storage phase, unauthorized entities cannot crack the ciphertext, thus ensuring data confidentiality.

[0050] Digital signature technology is used to further verify the authenticity and integrity of the data source. After encrypting the data, the data uploader uses their private key to generate a digital signature on the original user data. After the data is uploaded to the trusted data space, any data accessor or verifier can use the uploader's public key to verify the digital signature. If the verification is successful, it proves two points: first, the data was indeed provided by the corresponding uploader, the source is authentic and trustworthy, and there is no forged data; second, the data has not been tampered with during transmission and the content is complete. If the verification fails, it means that the data source is unknown or has been illegally tampered with, and the system will automatically intercept the data and not include it in the subsequent value assessment and scenario matching process.

[0051] Furthermore, the metadata corresponding to user data is written into a blockchain smart contract for permanent storage. The metadata includes information such as data type, generation time, data source, and signature digest, and is used to record the entire lifecycle information of the data. Each subject that needs to access the data has a unique identity on the blockchain. This identity is generated based on a public-private key mechanism and serves as the basis for identity verification and permission allocation. Permission management adopts role-based access control (RBAC) or attribute-based access control (ABAC). The subject's identity, role, business attributes, and other information are written into the smart contract, which automatically executes permission granting, modification, and revocation operations. When a subject initiates a data access request, the smart contract automatically verifies the subject's identity, corresponding permissions, and access requirements. If the requirements are met, the corresponding access permissions are granted; otherwise, the access request is directly rejected.

[0052] Leveraging the immutable nature of blockchain, all data access records and permission change records are permanently stored on the chain, forming a transparent and verifiable audit log. In the event of unauthorized access, abuse of permissions, or other violations, the on-chain logs can be used to quickly trace the operator, the time of operation, and the specific content of the operation, enabling security accountability.

[0053] This embodiment achieves end-to-end data confidentiality, tamper-proofing, and traceability through asymmetric encryption and digital signatures. Combined with blockchain smart contracts, it completes metadata storage and automated access control, ensuring the security and integrity of user data during transmission and storage, while also achieving refined and automated management of data access permissions.

[0054] In an optional implementation, the method further includes: Based on the overall revenue actually generated after matching data with business scenarios, and using the Shapley value and kernel function in cooperative game theory, the profit distribution share of each participating entity is calculated according to the marginal contribution of each participating entity to the overall revenue. The profit distribution share will be dynamically adjusted based on changes in the market environment, data circulation environment, and cooperative relationships among participating entities.

[0055] Specifically, after matching user data with business scenarios, actual business revenue is generated. This embodiment introduces a cooperative game mechanism, combining two core rules—Shapley value and kernel function—to achieve a fair distribution of revenue. The Shapley value is calculated based on the marginal contribution of each participating entity to the overall revenue, determining their corresponding revenue share. For example, the entities participating in revenue distribution include data providers, platform operators, technical support providers, and business implementers. These entities jointly participate in the entire process of data uploading, security protection, value assessment, scenario matching, and business implementation. The specific implementation of the Shapley value is as follows: iterate through all possible combinations of entity cooperation, calculate the marginal gain brought to the overall revenue when a single entity joins a certain combination, that is, the increase in the total revenue of the cooperation group after the entity joins. Summarize the marginal contribution of the entity under all cooperation combinations, calculate the average value, and the resulting value is the Shapley value corresponding to the entity, which is directly used as its basic revenue distribution share.

[0056] Furthermore, the kernel function defines a reasonable allocation range, or kernel region, ensuring that the final profit distribution scheme falls within this range. Through kernel function constraints, it is guaranteed that the profit received by each participating entity is not less than the minimum profit it could obtain individually through cooperation, thus ensuring the stability of the profit distribution scheme.

[0057] To adapt to changes in the external environment and internal cooperative relationships, a dynamic adjustment mechanism for profit distribution is established. The dynamic adjustment process follows three core principles to ensure that the adjustment process is fair, reasonable, and based on evidence: First, the principle of contribution, which uses the marginal contribution calculated by the Shapley value as the core basis, with higher contributions resulting in higher share of profits; second, the principle of fairness, which ensures that the entire distribution process is transparent and open, without favoring any particular participant and preventing excessive concentration of profits; and third, the principle of risk sharing, which adjusts the profit distribution ratio based on the input costs and risk tolerance of each participant to achieve a match between profits and risks.

[0058] The dynamic adjustment mechanism is mainly triggered by two types of key changes. The first type is changes in the market environment, including fluctuations in market demand, intensified industry competition, and adjustments to relevant policies, which will directly affect the overall revenue scale and the value of the main entities' contributions. The second type is changes in the data circulation environment, including changes in data processing efficiency, optimization of circulation paths, adjustments to security protection levels, and changes in the participation of main entities, which will change the actual contribution of each entity.

[0059] Dynamic adjustments can be implemented in two ways, without specific limitations here. The first is automatic adjustment based on preset rules. The system has built-in adjustment rules related to market demand, data circulation, and changes in contribution, and monitors various changing parameters in real time. When the trigger conditions are met, the allocation ratio is automatically calculated and updated. The second is corrective adjustment based on feedback. Reasonable feedback information from all participating entities is collected, and the revenue share is recalculated using a weighted average method. The adjustment results are executed through an automated system to ensure the efficiency and fairness of the adjustment.

[0060] This embodiment constructs a cooperative game-theoretic allocation model by combining Shapley values ​​and kernel functions. It can fairly divide the benefits according to the actual contributions of each subject. At the same time, the kernel function ensures the stability of the allocation scheme and eliminates the risk of cooperation breakdown. Coupled with a dynamic adjustment mechanism that adapts to changes in the market and data circulation, it takes into account both the fairness of allocation and the adaptability of the scenario. It can not only stimulate the cooperation of each participating subject, but also maintain long-term stable cooperation among multiple parties, making the entire data resource allocation system more fair, flexible and sustainable.

[0061] Secondly, embodiments of the present invention provide a data resource allocation device, see [link to relevant documentation]. Figure 2 This is a schematic diagram of an embodiment of a data resource allocation device provided by the present invention.

[0062] like Figure 2 As shown, the device includes: The data acquisition module 21 is used to acquire multimodal user data and upload it to a trusted data space; the multimodal user data includes, but is not limited to, user behavior logs, user attribute data, historical business response records, and user geographic location trajectories. The multi-dimensional evaluation module 22 is used to extract features from the user data from multiple dimensions and to obtain the initial value of the user data by weighted averaging of the feature values ​​of each dimension. Value prediction module 23 is used to input the feature values ​​of each dimension and the initial value into the pre-built data value prediction model, and output the potential commercial value of user data and the business scenario matching degree for each business scenario. The scenario matching module 24 is used to establish a bipartite graph of user data and business scenarios. Based on the potential business value and the matching degree of the business scenarios, it performs data matching with business scenarios in the bipartite graph and outputs the optimal matching dataset corresponding to each business scenario.

[0063] In one optional implementation, the process of constructing the data value prediction model includes: Obtain the multi-dimensional features and real value tags of historical user data; wherein, the real value tags include potential business value tags and business scenario matching tags; A regression analysis model was used to fit the linear relationship between each dimension feature and the real value label to obtain preliminary value fitting results. A decision tree model is used to learn nonlinear rules based on the features of each dimension and the preliminary value fitting results to obtain a single decision tree evaluation model. The random forest algorithm is used to integrate multiple decision tree models with different structures to obtain a data value prediction model.

[0064] In one optional implementation, the step of establishing a bipartite graph of user data and business scenarios, performing data-business scenario matching in the bipartite graph based on the potential business value and the matching degree of the business scenarios, and outputting the optimal matching dataset corresponding to each business scenario, includes: By using user data as data nodes and business scenarios as scenario nodes, a bipartite graph structure is constructed. A bipartite graph matching algorithm based on an attention mechanism is used to calculate the total matching value between each data node and each scene node; Using the maximization of the total matching value as the matching optimization objective, the optimal matching scheme of the bipartite graph is solved; Based on the optimal matching scheme, output the optimal matching data list corresponding to each business scenario.

[0065] In one alternative implementation, the total matching value is the product of the potential business value and the business scenario matching degree.

[0066] In an optional embodiment, the device is further configured to: During the process of uploading the multimodal user data to the trusted data space, an asymmetric encryption algorithm is used to encrypt and protect the data, and digital signature technology is used to verify the authenticity of the data source and the integrity of the data. The metadata of the user data, including data type, generation time, data source, signature digest, and other information, is written into a blockchain smart contract for storage.

[0067] In an optional embodiment, the device is further configured to: Based on the overall revenue actually generated after matching data with business scenarios, and using the Shapley value and kernel function in cooperative game theory, the profit distribution share of each participating entity is calculated according to the marginal contribution of each participating entity to the overall revenue. The profit distribution share will be dynamically adjusted based on changes in the market environment, data circulation environment, and cooperative relationships among participating entities.

[0068] It should be noted that the data resource allocation device provided in this embodiment of the invention is used to execute all the process steps of the data resource allocation method in the above embodiment. The working principle and beneficial effects of the two are one-to-one, so they will not be described again.

[0069] Thirdly, embodiments of the present invention provide an electronic device, see [link to previous document]. Figure 3 The diagram shown is a structural schematic of an electronic device provided in an embodiment of the present invention.

[0070] like Figure 3 As shown, the device includes: Memory 31 is used to store computer programs; Processor 32 is used to execute the computer program; When the processor 32 executes the computer program, it implements the data resource allocation method as described in any of the above embodiments.

[0071] For example, the computer program may be divided into one or more modules / units, which are stored in the memory 31 and executed by the processor 32 to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0072] The processor 32 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0073] The memory 31 can be used to store the computer programs and / or modules. The processor 32 implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory 31 and calling the data stored in the memory 31. The memory 31 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory 31 may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital card (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0074] It should be noted that the aforementioned electronic devices include, but are not limited to, processors and memory, as will be understood by those skilled in the art. Figure 3 The structural diagram is merely an example of the electronic device described above and does not constitute a limitation on the electronic device. It may include more components than shown in the diagram, or combine certain components, or use different components.

[0075] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, which, when executed, implements the data resource allocation method described in any of the above embodiments.

[0076] It should be understood that the implementation of all or part of the above-described data resource allocation method can also be accomplished by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the above-described data resource allocation method. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0077] Fifthly, embodiments of this application also provide a computer program product, which is stored in a storage medium and executed by at least one processor to implement the data resource allocation method described in any of the above embodiments.

[0078] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. It should be noted that, for those skilled in the art, several equivalent obvious modifications and / or equivalent substitutions can be made without departing from the technical principles of the present invention, and these obvious modifications and / or equivalent substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A data resource allocation method, characterized in that, include: Acquire multimodal user data and upload it to a trusted data space; The multimodal user data includes, but is not limited to, user behavior logs, user attribute data, historical business response records, and user geographic location trajectories; The user data is feature extracted from multiple dimensions, and the initial value of the user data is obtained by weighted averaging of the feature values ​​of each dimension. The feature values ​​of each dimension and the initial value are input into a pre-built data value prediction model, which outputs the potential commercial value of user data and the business scenario matching degree for each business scenario. Establish a bipartite graph of user data and business scenarios. Based on the potential business value and the matching degree of the business scenarios, perform data matching with business scenarios in the bipartite graph, and output the optimal matching dataset corresponding to each business scenario.

2. The data resource allocation method as described in claim 1, characterized in that, The construction process of the data value prediction model includes: Obtain the multi-dimensional features and real value tags of historical user data; wherein, the real value tags include potential business value tags and business scenario matching tags; A regression analysis model was used to fit the linear relationship between each dimension feature and the real value label to obtain preliminary value fitting results. A decision tree model is used to learn nonlinear rules based on the features of each dimension and the preliminary value fitting results to obtain a single decision tree evaluation model. The random forest algorithm is used to integrate multiple decision tree models with different structures to obtain a data value prediction model.

3. The data resource allocation method as described in claim 1, characterized in that, The process of establishing a bipartite graph of user data and business scenarios involves matching data with business scenarios within the bipartite graph based on the potential business value and the matching degree of the business scenarios, and outputting the optimal matching dataset for each business scenario, including: By using user data as data nodes and business scenarios as scenario nodes, a bipartite graph structure is constructed. A bipartite graph matching algorithm based on an attention mechanism is used to calculate the total matching value between each data node and each scene node; Using the maximization of the total matching value as the matching optimization objective, the optimal matching scheme of the bipartite graph is solved; Based on the optimal matching scheme, output the optimal matching data list corresponding to each business scenario.

4. The data resource allocation method as described in claim 3, characterized in that, The total matching value is the product of the potential business value and the matching degree of the business scenario.

5. The data resource allocation method as described in claim 1, characterized in that, The method further includes: During the process of uploading the multimodal user data to the trusted data space, an asymmetric encryption algorithm is used to encrypt and protect the data, and digital signature technology is used to verify the authenticity of the data source and the integrity of the data. The metadata of the user data, including data type, generation time, data source, signature digest, and other information, is written into a blockchain smart contract for storage.

6. The data resource allocation method as described in claim 1, characterized in that, The method further includes: Based on the overall revenue actually generated after matching data with business scenarios, and using the Shapley value and kernel function in cooperative game theory, the profit distribution share of each participating entity is calculated according to the marginal contribution of each participating entity to the overall revenue. The profit distribution share will be dynamically adjusted based on changes in the market environment, data flow environment, and cooperative relationships among participating entities.

7. A data resource allocation device, characterized in that, include: The data acquisition module is used to acquire multimodal user data and upload it to a trusted data space. The multimodal user data includes, but is not limited to, user behavior logs, user attribute data, historical business response records, and user geographic location trajectories; The multi-dimensional evaluation module is used to extract features from the user data from multiple dimensions and to obtain the initial value of the user data by weighted averaging of the feature values ​​of each dimension. The value prediction module is used to input the feature values ​​of each dimension and the initial value into the pre-built data value prediction model, and output the potential commercial value of user data and the business scenario matching degree for each business scenario. The scenario matching module is used to establish a bipartite graph of user data and business scenarios. Based on the potential business value and the matching degree of the business scenarios, it performs data matching with business scenarios in the bipartite graph and outputs the optimal matching dataset corresponding to each business scenario.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program; Wherein, when the processor executes the computer program, it implements the data resource allocation method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed, implements the data resource allocation method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes computer instructions that, when executed by a processor, implement the data resource allocation method as described in any one of claims 1 to 6.

Images