Data value evaluation method and device, computer equipment and storage medium
By performing feature analysis and transfer learning on low-frequency trading data, combined with graph neural network and pre-trained value evaluation model, the problem of deviation in low-frequency trading data value evaluation results is solved, and the accurate quantification of the value of low-frequency trading data is achieved.
Patent Information
- Application Number
- CN202411640172.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-15
- Publication Date
- 2025-03-21
AI Technical Summary
When conducting low-frequency trading data value evaluation, due to the small amount of data and limited information coverage, it is difficult to effectively capture the overall picture of user behavior and market dynamic changes, resulting in large deviations in the evaluation results.
By obtaining transaction data and transaction frequency impact factors, a data relationship map of high-frequency transaction data is constructed, and an implicit correlation characteristics between users and commodities are extracted using graph neural network algorithm. Then, these implicit features are transferred to low-frequency trading data through transfer learning, enhancing their feature representation capabilities, and ultimately inputting low-frequency data features into the pre-trained value evaluation model to obtain more accurate evaluation results.
It effectively alleviates the problem of insufficient coverage of low-frequency data, reduces evaluation bias, and realizes accurate quantification of the value of low-frequency trading data.
Smart Images

Figure CN119669891A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of artificial intelligence technology, and specifically relates to a data value assessment method, device, computer equipment and storage medium. Background Art
[0002] The value of data is reflected in many aspects and is of great significance to enterprises, governments and individuals. In the era of digital economy, a comprehensive understanding of data and its value, as well as the effective management and application of data resources, will be the key to innovative development. For example, enterprises and organizations can use data to understand market trends, consumer demand and competitor behavior, so as to formulate more effective marketing strategies and business plans; for another example, the government can understand the social development status and people's livelihood needs by analyzing population data, traffic data and environmental data, and formulate more scientific and reasonable policies and plans.
[0003] At present, when evaluating the value of transaction data, for high-frequency transaction data, since it contains rich information on user behavior patterns and market trends, a series of methods such as data mining and evaluation can be used to obtain relatively accurate data value evaluation results. However, for low-frequency transaction data, due to its relatively small amount of data and limited information coverage, it is difficult to effectively capture the full picture of user behavior and market dynamics, resulting in a large deviation in the evaluation results. At the same time, low-frequency transaction data is susceptible to noise interference, and the data quality is uneven, which further increases the difficulty and uncertainty of value evaluation. Summary of the invention
[0004] The purpose of the embodiments of the present application is to propose a data value assessment method, device, computer equipment and storage medium to solve the technical problem that when conducting value assessment on low-frequency transaction data, due to its relatively small amount of data and limited information coverage, it is difficult to effectively capture the full picture of user behavior and market dynamics, resulting in large deviations in the assessment results.
[0005] In order to solve the above technical problems, the present application embodiment provides a data value assessment method, which adopts the following technical solution:
[0006] A data value assessment method, comprising:
[0007] Obtain transaction data and transaction frequency influencing factors, classify transaction data according to the transaction frequency influencing factors, and obtain high-frequency transaction data and low-frequency transaction data;
[0008] Perform feature analysis on high-frequency trading data to obtain the features of high-frequency trading data;
[0009] Based on the characteristics of high-frequency transaction data, a data relationship graph of high-frequency transaction data is constructed, where the nodes of the data relationship graph represent users or commodities, and the edges of the data relationship graph represent transaction relationships;
[0010] Use graph neural network algorithm to traverse the data relationship graph to obtain the implicit correlation features between users and products;
[0011] Based on the implicit correlation features, the transfer learning algorithm is used to analyze the features of low-frequency transaction data to obtain the features of low-frequency transaction data;
[0012] Import the low-frequency transaction data features into the pre-trained data value assessment model to obtain the value assessment results of the low-frequency transaction data.
[0013] Further, the steps of obtaining transaction data and transaction frequency influencing factors, classifying the transaction data according to the transaction frequency influencing factors, and obtaining high-frequency transaction data and low-frequency transaction data specifically include:
[0014] Obtain transaction data and clean the transaction data to obtain standardized transaction data;
[0015] According to the data classification requirements, the matching transaction frequency impact factor is obtained, and the transaction frequency of each transaction data is calculated based on the transaction frequency impact factor;
[0016] The transaction frequency of each transaction data is compared with the preset transaction frequency threshold to obtain high-frequency transaction data and low-frequency transaction data.
[0017] Furthermore, based on the characteristics of high-frequency trading data, the step of constructing a data relationship graph of high-frequency trading data specifically includes:
[0018] Classify the high-frequency transaction data features to obtain user data features, commodity data features, and transaction relationship features;
[0019] Initialize a graph object and create graph nodes in the graph object based on user data features and product data features;
[0020] Calculate the edge weights, and generate the edges of the data relationship graph based on the edge weights and transaction relationship characteristics;
[0021] The edges of the data relationship graph are embedded into the graph object to obtain the data relationship graph of high-frequency trading data.
[0022] Furthermore, the steps of calculating the edge weights and generating the edges of the data relationship graph based on the edge weights and transaction relationship characteristics specifically include:
[0023] Identify transaction frequency and transaction amount in high-frequency trading data;
[0024] Calculate the weight value of the transaction frequency to obtain a first weight value;
[0025] Calculate the weight value of the transaction amount to obtain a second weight value;
[0026] The first weight value and the second weight value are weighted to obtain the weight of the edge;
[0027] The transaction relationship features are weighted using the edge weights, and the edges of the data relationship graph are drawn based on the weighted transaction relationship features.
[0028] Furthermore, the graph neural network algorithm is used to perform a graph traversal operation on the data relationship graph to obtain the implicit correlation features between users and products, which specifically includes:
[0029] Convert the format of the relationship graph into graph data format;
[0030] Import the relationship graph converted into a graph data format into a preset initial feature matrix to obtain a graph feature matrix;
[0031] The graph feature matrix is input into the pre-trained graph neural network, and the convolution layer of the graph neural network is used to traverse and convolve the graph feature matrix to obtain implicit correlation features.
[0032] Furthermore, the graph neural network is calculated based on the graph neural network algorithm. The graph neural network includes a first convolution layer and a second convolution layer. The first convolution layer is used to convolve the user data features, and the second convolution layer is used to convolve the product data features. The convolution layer of the graph neural network is used to traverse and convolve the graph feature matrix to obtain the implicit correlation feature step, which specifically includes:
[0033] Use the preset sliding window to traverse the graph feature matrix and identify the user data feature submatrix and the product data feature submatrix;
[0034] Use the first convolution layer to perform a convolution operation on the user data feature submatrix to obtain a first convolution feature;
[0035] Use the second convolution layer to perform a convolution operation on the product data feature submatrix to obtain the second convolution feature;
[0036] The first convolution feature and the second convolution feature are fused to obtain implicit correlation features.
[0037] Furthermore, based on the implicit correlation features, the low-frequency transaction data is analyzed using a transfer learning algorithm to obtain the features of the low-frequency transaction data, specifically including:
[0038] Determine the data type label that matches the low-frequency transaction data and obtain the target data type label;
[0039] Determine the high-frequency trading data matching the target data type label, and obtain the target high-frequency trading data;
[0040] Obtaining implicit correlation features corresponding to target high-frequency trading data to obtain target implicit correlation features;
[0041] Extract features of low-frequency transaction data to obtain initial data features;
[0042] The initial data features and the target implicit correlation features are fused to obtain the frequent transaction data features.
[0043] In order to solve the above technical problems, the embodiment of the present application also provides a data value assessment device, which adopts the following technical solution:
[0044] A data value assessment device, comprising:
[0045] The data classification module is used to obtain transaction data and transaction frequency influencing factors, and classify the transaction data according to the transaction frequency influencing factors to obtain high-frequency transaction data and low-frequency transaction data;
[0046] High-frequency analysis module, used to perform feature analysis on high-frequency trading data and obtain the features of high-frequency trading data;
[0047] A graph construction module is used to construct a data relationship graph of high-frequency transaction data based on the characteristics of high-frequency transaction data, wherein the nodes of the data relationship graph represent users or commodities, and the edges of the data relationship graph represent transaction relationships;
[0048] The graph traversal module is used to perform graph traversal operations on the data relationship graph using a graph neural network algorithm to obtain implicit association features between users and products;
[0049] The low-frequency analysis module is used to perform feature analysis on low-frequency transaction data based on implicit correlation features using a transfer learning algorithm to obtain low-frequency transaction data features;
[0050] The value assessment module is used to import the low-frequency transaction data features into the pre-trained data value assessment model to obtain the value assessment results of the low-frequency transaction data.
[0051] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:
[0052] A computer device comprises a memory and a processor, wherein the memory stores computer-readable instructions, and the processor implements the steps of any of the above-mentioned data value assessment methods when executing the computer-readable instructions.
[0053] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:
[0054] A computer-readable storage medium having computer-readable instructions stored thereon, wherein the computer-readable instructions, when executed by a processor, implement the steps of the data value assessment method as described in any one of the above.
[0055] Compared with the prior art, the embodiments of the present application have the following beneficial effects:
[0056] The present application discloses a data value assessment method, device, computer equipment and storage medium, which belongs to the field of artificial intelligence technology. The present application first classifies transaction data based on transaction frequency to separate high-frequency and low-frequency transaction data. Then, by constructing a data relationship graph for high-frequency transaction data and traversing the graph using a graph neural network algorithm, the implicit correlation features between users and commodities are extracted. The implicit features in these high-frequency transaction data reflect user behavior and market dynamics. Afterwards, through transfer learning, these implicit features are migrated to low-frequency transaction data, so that low-frequency data can have richer feature representations under limited coverage and capture deep information about user behavior. Finally, the low-frequency data features are input into a pre-trained value assessment model to obtain a more accurate assessment result. The present application effectively alleviates the problem of insufficient coverage of low-frequency data, reduces assessment bias, and realizes accurate quantification of the value of low-frequency transaction data. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0058] Figure 1 An exemplary system architecture diagram is shown in which the present application can be applied;
[0059] Figure 2 A flow chart showing an embodiment of a method for evaluating data value according to the present application is shown;
[0060] Figure 3 A schematic diagram showing the structure of an embodiment of a data value assessment device according to the present application is shown;
[0061] Figure 4 A schematic structural diagram of an embodiment of a computer device according to the present application is shown. DETAILED DESCRIPTION
[0062] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.
[0063] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.
[0064] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.
[0065] like Figure 1 As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables.
[0066] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.
[0067] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to a laptop computer 1011, a tablet computer 1012 or a mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.
[0068] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .
[0069] It should be noted that the data value assessment method provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the data value assessment device is generally arranged in the server / terminal device.
[0070] It should be understood that Figure 1 The numbers of terminal devices, networks and servers in the embodiment are merely illustrative. The above system may have any number of terminal devices, networks and servers according to implementation requirements.
[0071] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for data value assessment according to the present application. The data value assessment method comprises the following steps:
[0072] S201, acquiring transaction data and transaction frequency influencing factors, and classifying the transaction data according to the transaction frequency influencing factors to obtain high-frequency transaction data and low-frequency transaction data.
[0073] Specifically, in order to effectively utilize the different characteristics of high-frequency and low-frequency transaction data, we first obtain all transaction records from the original transaction data set and introduce the frequency influencing factor. Then, by setting a frequency threshold, we divide the transaction data into high-frequency and low-frequency categories. The frequency above the threshold is high-frequency transaction data, and vice versa. High-frequency transaction data usually contains high-frequency interaction information between users and commodities, representing more stable user behavior patterns and market dynamics; while low-frequency transaction data is difficult to directly reflect the core preferences of users due to its low frequency, but it still plays an important supplementary role in the overall data value assessment.
[0074] The frequency impact factor can be obtained through multi-dimensional statistics such as the number of transactions, active time periods, and commodity turnover rate of each user or commodity, reflecting the activity of transactions. This factor combines information such as the number of transactions, transaction time intervals, and the depth of interaction between users or commodities, and is used to classify high-frequency and low-frequency transaction data. In data value assessment, the transaction frequency impact factor helps distinguish high-activity and low-activity data, making the analysis more targeted, thereby better capturing user behavior patterns and market dynamics.
[0075] S202, performing feature analysis on high-frequency trading data to obtain features of the high-frequency trading data.
[0076] Specifically, more representative data features between users and commodities are extracted from high-frequency transaction data. Since high-frequency data contains repeated behaviors of users and multiple circulations of commodities, it can provide a variety of information such as behavioral preferences, consumption capacity, and time trends. First, based on indicators such as user purchase frequency, purchase cycle, and total consumption, characteristics such as user preference intensity, purchase cycle characteristics, and price sensitivity can be extracted. Secondly, through the number of times a commodity is purchased, the repeat purchase rate, and the hot-selling period, the popularity, life cycle, and consumption scenario of the commodity can be extracted. High-frequency data can also reflect the potential group pattern of user behavior. Therefore, it can be combined with methods such as cluster analysis to further summarize the characteristics in high-frequency transactions into characteristic labels for various types of users or commodities.
[0077] After obtaining the features of high-frequency trading data, the features of high-frequency trading data can be imported into the pre-trained data value assessment model to obtain the value assessment results of the high-frequency trading data. The data value assessment model is a pre-trained comprehensive data value assessment model that combines multiple data features, historical trading behaviors and market dynamics. It is trained through machine learning algorithms (such as deep learning, regression analysis, etc.) and can automatically identify the relationship between different data features and data value. The model has been trained on a large amount of high-frequency trading data and can accurately assess the commercial value, potential benefits and market influence of the data. When evaluating high-frequency trading data, the model will generate a corresponding value score based on the input feature data (such as user behavior, product popularity, etc.) to help decision makers understand the real market potential of the data.
[0078] S203, constructing a data relationship graph of the high-frequency transaction data based on the characteristics of the high-frequency transaction data, wherein the nodes of the data relationship graph represent users or commodities, and the edges of the data relationship graph represent transaction relationships.
[0079] Specifically, the construction of the data relationship graph mainly converts the characteristics of users, commodities and transaction relationships into a graph structure. The nodes of the graph are users and commodities, and the edges represent the transaction relationship between users and commodities. For each transaction record, if the user has a purchase behavior for the commodity, a corresponding edge is created in the graph and a weight is assigned. The weight can be the transaction frequency, amount or purchase period, etc. The specific weight selection depends on the analysis requirements. The graph constructed in this way not only shows the direct transaction relationship between users and commodities, but also reflects the indirect relationship between users and users, and commodities and commodities. In the specific implementation, graph processing tools such as NetworkX or DGL can be used to construct the graph and store it in a format that is easy to operate. The data relationship graph structure will provide support for the subsequent use of graph neural networks to analyze the implicit association characteristics between users and commodities, and help capture more complex association patterns.
[0080] S204: Use a graph neural network algorithm to perform a graph traversal operation on the data relationship graph to obtain implicit correlation features between users and products.
[0081] Graph Neural Network (GNN) learns the features of user and product nodes through an iterative information transmission mechanism to obtain potential correlation features in the graph. Graph Neural Network uses neural networks to learn graph structure data, extract and discover features and patterns in graph structure data, and meet the needs of graph learning tasks such as clustering, classification, prediction, segmentation, and generation. Entities in graph structure data are represented in the form of nodes, and the relationships between entities are represented in the form of edges. The goal of GNN is to learn useful representations from graph structure data and use these representations for various tasks, such as node classification, graph classification, link prediction, etc.
[0082] In the specific implementation, classic GNN algorithms such as GCN and GraphSAGE can be used to perform graph convolution operations on the graph, so that the features of the nodes aggregate the information from the neighboring nodes layer by layer. After multiple layers of convolution, GNN can capture the complex relationship between multi-hop neighbors and generate implicit association features of users and products. These implicit features not only include the individual information of users and products, but also reflect the characteristics of their surrounding neighbors, which is a more comprehensive expression of user purchase preferences and product trends. During the training process, the GNN model can use self-supervised learning to optimize the model through tasks such as node feature reconstruction and edge prediction, so that the final extracted implicit association features are more effective.
[0083] S205, based on the implicit correlation features, use the transfer learning algorithm to perform feature analysis on the low-frequency transaction data to obtain the features of the low-frequency transaction data.
[0084] Transfer learning is used to transfer implicit features learned from high-frequency trading data to low-frequency trading data, thereby making up for the lack of information in low-frequency data. The transfer learning algorithm uses the knowledge learned on the source task to help the learning of the target task. It assumes that there is a certain similarity between the source task and the target task, so the learning efficiency of the target task can be improved by transferring the knowledge on the source task. The key to transfer learning is to find the common points between the source task and the target task and effectively transfer the knowledge on the source task to the target task.
[0085] In the specific implementation, the system uses a pre-trained GNN model to transfer the implicit features of high-frequency data as knowledge to low-frequency data, and adjusts the model through transfer learning algorithms such as Fine-tuning or Domain Adaptation to adapt the model to the distribution of low-frequency data. The migrated model will extract features similar to those of high-frequency data on low-frequency transaction data to ensure that low-frequency data can also have the same feature richness. In this way, although the transaction volume of low-frequency data itself is small, it can also reflect core information such as user preferences and product popularity with the support of implicit correlation features, improving the feature representation effect of low-frequency transaction data.
[0086] S206, importing the low-frequency transaction data features into the pre-trained data value assessment model to obtain the value assessment results of the low-frequency transaction data.
[0087] Specifically, the features of low-frequency trading data are input into the pre-trained value assessment model to generate the value assessment results of the low-frequency data. The value assessment model is usually built based on machine learning algorithms, such as regression models or neural network models. The training process combines the features of historical high-frequency trading data with the real value labels, so that the model can learn how to accurately estimate the actual value of the data from the data features. By importing the features of low-frequency trading data into the model, the assessment model can generate a value score or grade based on the input features, quantify the actual value of the low-frequency trading data, and output the results.
[0088] In the above embodiment, the present application first classifies the transaction data based on the transaction frequency, and separates the high-frequency and low-frequency transaction data. Then, by constructing a data relationship graph for the high-frequency transaction data, and using a graph neural network algorithm to traverse the graph, the implicit correlation features between users and commodities are extracted. The implicit features in these high-frequency transaction data reflect user behavior and market dynamics. Afterwards, through transfer learning, these implicit features are transferred to low-frequency transaction data, so that low-frequency data can have richer feature representations under limited coverage and capture deep information about user behavior. Finally, the low-frequency data features are input into the pre-trained value assessment model to obtain a more accurate assessment result. The present application effectively alleviates the problem of insufficient coverage of low-frequency data, reduces assessment bias, and realizes accurate quantification of the value of low-frequency transaction data.
[0089] Further, the steps of obtaining transaction data and transaction frequency influencing factors, classifying the transaction data according to the transaction frequency influencing factors, and obtaining high-frequency transaction data and low-frequency transaction data specifically include:
[0090] Obtain transaction data and clean the transaction data to obtain standardized transaction data;
[0091] According to the data classification requirements, the matching transaction frequency impact factor is obtained, and the transaction frequency of each transaction data is calculated based on the transaction frequency impact factor;
[0092] The transaction frequency of each transaction data is compared with the preset transaction frequency threshold to obtain high-frequency transaction data and low-frequency transaction data.
[0093] In this embodiment, a complete transaction data set is first obtained from the original data source, and the data set is cleaned. The cleaning operation includes deduplication, missing value completion, outlier detection and processing, etc., to ensure data accuracy and consistency, and standardize the data. In the standardized transaction data after cleaning, according to the data classification requirements, select the appropriate transaction frequency influencing factor. The transaction frequency influencing factor can include indicators such as the number of transactions, transaction cycle, user's active period, and commodity repurchase rate. The specific selection is determined according to business needs. Based on this influencing factor, the transaction frequency of each transaction record is calculated, and the frequency value is compared with the set threshold to classify high-frequency and low-frequency transaction data. High-frequency transaction data reflects frequent user-commodity interactions and helps capture user behavior patterns; while low-frequency transaction data represents relatively scattered behaviors, but contains special transaction characteristics.
[0094] Through the above steps, the cleaned data is efficiently divided into high-frequency and low-frequency transaction data, which improves the accuracy and reliability of feature analysis and model evaluation.
[0095] Furthermore, based on the characteristics of high-frequency trading data, the step of constructing a data relationship graph of high-frequency trading data specifically includes:
[0096] Classify the high-frequency transaction data features to obtain user data features, commodity data features, and transaction relationship features;
[0097] Initialize a graph object and create graph nodes in the graph object based on user data features and product data features;
[0098] Calculate the edge weights, and generate the edges of the data relationship graph based on the edge weights and transaction relationship characteristics;
[0099] The edges of the data relationship graph are embedded into the graph object to obtain the data relationship graph of high-frequency trading data.
[0100] In this embodiment, the features of high-frequency transaction data are first finely classified into user data features, commodity data features, and transaction relationship features. User data features include basic attributes of users (such as age, gender, and region) and behavioral characteristics (such as activity and consumption preferences); commodity data features include commodity categories, price ranges, sales cycles, etc.; transaction relationship features reflect the specific transaction situation between users and commodities, including transaction frequency, amount, etc. Then, a graph object is initialized to represent the user-commodity relationship graph, and each item in the user and commodity data features is converted into a node in the graph to distinguish different types of nodes. Then, the weight of the edge in the graph is calculated according to the transaction relationship features. For example, the transaction frequency or transaction amount can be used as the weight of the edge to indicate the intensity or frequency of the transaction. The generated edge connects the user and commodity nodes, and after being embedded in the graph object, a complete relationship graph is formed, so that the multi-dimensional association information of users and commodities can be presented in a structured manner. This relationship graph can not only reveal the direct transaction relationship between users and commodities, but also further explore deeper indirect relationships and association patterns.
[0101] Through the above steps, a relationship map of high-frequency trading data is constructed, and the multi-level relationships between data are visualized and structured in order to mine implicit user-product association features.
[0102] Furthermore, the steps of calculating the edge weights and generating the edges of the data relationship graph based on the edge weights and transaction relationship characteristics specifically include:
[0103] Identify transaction frequency and transaction amount in high-frequency trading data;
[0104] Calculate the weight value of the transaction frequency to obtain a first weight value;
[0105] Calculate the weight value of the transaction amount to obtain a second weight value;
[0106] The first weight value and the second weight value are weighted to obtain the weight of the edge;
[0107] The transaction relationship features are weighted using the edge weights, and the edges of the data relationship graph are drawn based on the weighted transaction relationship features.
[0108] In this embodiment, two key features, transaction frequency and transaction amount, are first extracted from high-frequency transaction data. Transaction frequency reflects the frequency of interaction between a certain user and a commodity, while transaction amount indicates the value of each transaction. Next, by calculating the weight value of transaction frequency, a first weight value is obtained, which reflects the interaction intensity between the user and the commodity. Subsequently, the weight value of transaction amount is calculated to obtain a second weight value, which reflects the economic importance of the transaction. Next, the first weight value and the second weight value are weighted and summed to obtain a comprehensive edge weight, which represents the comprehensive transaction relationship intensity between the user and the commodity. Finally, based on this comprehensive weight, the transaction relationship features are weighted, and the weighted transaction relationship features can more accurately represent the actual influence of the transaction. Finally, these weighted transaction relationship features are used to draw the edges of the data relationship map to ensure that the edges in the map can truly reflect the actual interaction intensity and economic influence between the user and the commodity.
[0109] For example, suppose user A has conducted three transactions with commodity X, 10 times, 5 times, and 3 times, respectively, and the transaction amounts of each transaction are 100 yuan, 200 yuan, and 50 yuan, respectively. When calculating the weight value of transaction frequency, it can be evaluated according to the number of transactions between the user and the commodity. Records with higher transaction frequency will be given a larger weight. For example, the weight of the first transaction is 10, the second is 5, and the third is 3. Then, calculate the weight value of transaction amount. Records with larger transaction amounts can be given a higher weight. For example, the amount of the first transaction is 100 yuan, the weight is 100, the amount of the second transaction is 200 yuan, the weight is 200, and the weight of the third transaction is 50 yuan. Next, add the frequency weight and the amount weight of each transaction to obtain the comprehensive edge weight. For example, the comprehensive weight of the first transaction is 10 (frequency) + 100 (amount) = 110, the second transaction is 5 + 200 = 205, and the third transaction is 3 + 50 = 53. Ultimately, the edges drawn based on these weights will accurately reflect the interaction intensity and economic impact between each user and item.
[0110] Through the above steps, edge weight calculation and empowerment can more accurately characterize the transaction relationship between users and products, and improve the authenticity and reliability of the relationship graph.
[0111] Furthermore, the graph neural network algorithm is used to perform a graph traversal operation on the data relationship graph to obtain the implicit correlation features between users and products, which specifically includes:
[0112] Convert the format of the relationship graph into graph data format;
[0113] Import the relationship graph converted into a graph data format into a preset initial feature matrix to obtain a graph feature matrix;
[0114] The graph feature matrix is input into the pre-trained graph neural network, and the convolution layer of the graph neural network is used to traverse and convolve the graph feature matrix to obtain implicit correlation features.
[0115] In this embodiment, the constructed data relationship graph is first converted into a graph data format to ensure that it meets the requirements of graph neural network (GNN) processing. Specifically, the nodes and edges in the graph are converted into the adjacency matrix and node feature matrix of the graph to facilitate the processing of the graph neural network. Next, the converted graph data format is imported into the preset initial feature matrix, which contains preliminary feature information of users and commodities, such as user consumption preferences, popularity of commodities, etc. Then, the graph feature matrix is input into the pre-trained graph neural network model. The graph neural network performs multi-layer traversal convolution operations on the graph feature matrix through its convolution layer. In each layer, the node will exchange and fuse information with its neighboring nodes, thereby gradually learning deeper implicit association features. Ultimately, the implicit features obtained through the convolution operation can capture the complex relationship between users and commodities, such as the potential preferences of users or the cross-selling potential of commodities. These implicit features will provide key support for low-frequency data feature analysis and value assessment.
[0116] Through the above steps, the graph neural network effectively extracts the deep-level correlation features implicit in the data relationship graph, provides more expressive feature representation for subsequent data analysis and value assessment, and improves the accuracy and reliability of the model.
[0117] Furthermore, the graph neural network is calculated based on the graph neural network algorithm. The graph neural network includes a first convolution layer and a second convolution layer. The first convolution layer is used to convolve the user data features, and the second convolution layer is used to convolve the product data features. The convolution layer of the graph neural network is used to traverse and convolve the graph feature matrix to obtain the implicit correlation feature step, which specifically includes:
[0118] Use the preset sliding window to traverse the graph feature matrix and identify the user data feature submatrix and the product data feature submatrix;
[0119] Use the first convolution layer to perform a convolution operation on the user data feature submatrix to obtain a first convolution feature;
[0120] Use the second convolution layer to perform a convolution operation on the product data feature submatrix to obtain the second convolution feature;
[0121] The first convolution feature and the second convolution feature are fused to obtain implicit correlation features.
[0122] In this embodiment, first, a preset sliding window is used to perform a traversal operation in the graph feature matrix. The size and step size of the sliding window can be set according to specific needs. Through this sliding window operation, the submatrix containing user data features and commodity data features can be effectively identified, corresponding to the node information of the user and the commodity respectively. Then, the first convolution layer is used to perform a convolution operation on the user data feature submatrix. The convolution process can extract the key features of user behavior by weighted averaging the information of the node and its neighboring nodes. For example, the convolution operation can help identify the user's preference for different commodity categories. Then, the second convolution layer is used to perform a convolution operation on the commodity data feature submatrix to extract commodity features, such as the popularity of the commodity, price fluctuations, etc. The two convolution layers perform feature extraction for user and commodity data respectively. In this way, the features of the user and the commodity are fully captured and expressed. Finally, the first convolution feature and the second convolution feature are feature fused, and the implicit association features between the user and the commodity are obtained by splicing, weighting or other fusion methods. These implicit features can reveal more complex non-explicit relationships between users and commodities.
[0123] For example, suppose there is an e-commerce platform where there is a transaction relationship between user A and product X. The features of user A include purchase history, browsing behavior, etc., and the features of product X include price, sales volume, etc. First, use the sliding window operation to extract the feature submatrix of user A and the feature submatrix of product X from the graph feature matrix. Then, the first convolutional layer performs a convolution operation on the feature submatrix of user A, which may fuse the preference information of user A with the historical transaction products to obtain the user's behavior pattern characteristics. For example, the feature that user A prefers high-priced products. Then, the second convolutional layer convolves the feature submatrix of product X to extract the sales trend and price sensitivity of product X. Finally, the convolutional features of user A and product X are fused to obtain implicit association features, such as user A's potential purchase intention for product X. This association feature helps further predict the user's possible purchase behavior.
[0124] Through the above steps, the graph neural network effectively extracts the complex implicit correlation features between users and products through multi-layer convolution operations, improves the model's ability to recognize deep user behavior patterns, and enhances the accuracy of value assessment.
[0125] Furthermore, based on the implicit correlation features, the low-frequency transaction data is analyzed using a transfer learning algorithm to obtain the features of the low-frequency transaction data, specifically including:
[0126] Determine the data type label that matches the low-frequency transaction data and obtain the target data type label;
[0127] Determine the high-frequency trading data matching the target data type label, and obtain the target high-frequency trading data;
[0128] Obtaining implicit correlation features corresponding to target high-frequency trading data to obtain target implicit correlation features;
[0129] Extract features of low-frequency transaction data to obtain initial data features;
[0130] The initial data features and the target implicit correlation features are fused to obtain the frequent transaction data features.
[0131] In this embodiment, first, by analyzing the characteristics of low-frequency transaction data, the data type label that matches the low-frequency transaction data is determined. The target data type label can be set according to business needs, such as "high-value goods" or "potential high-frequency users". Then, based on the target data type label, the data that matches it is found from the high-frequency transaction data to obtain the target high-frequency transaction data. For example, if the low-frequency transaction data corresponds to a potential high-frequency user, then the user data that exhibits similar behavior in high-frequency transactions can be selected. Next, the implicit association features corresponding to the target high-frequency transaction data are obtained. These features are extracted from the high-frequency transaction data through methods such as graph neural networks to reveal the potential association pattern between users and commodities. Then, the low-frequency transaction data is feature extracted to obtain initial data features. These initial features can reflect the basic information of the low-frequency transaction data, but may not fully capture the deep-level behavior of the user. Finally, the initial features of the low-frequency transaction data are fused with the target implicit association features, and the feature expression ability of the low-frequency data is improved through feature fusion, making it more potential. This method enhances the representation ability of low-frequency data by leveraging the rich features of high-frequency transaction data, thereby improving the effect of low-frequency transaction data analysis.
[0132] Through the above steps, transfer learning effectively transfers the knowledge in high-frequency trading data to low-frequency trading data, making up for the problem of insufficient low-frequency data samples, improving the feature analysis ability of low-frequency data, and enhancing the model's evaluation accuracy of low-frequency data.
[0133] In the above embodiments, the present application discloses a data value assessment method, which belongs to the field of artificial intelligence technology. The present application first classifies transaction data based on transaction frequency to separate high-frequency and low-frequency transaction data. Then, by constructing a data relationship graph for high-frequency transaction data and traversing the graph using a graph neural network algorithm, the implicit correlation features between users and commodities are extracted. The implicit features in these high-frequency transaction data reflect user behavior and market dynamics. Afterwards, through transfer learning, these implicit features are migrated to low-frequency transaction data, so that low-frequency data can have richer feature representations under limited coverage and capture deep information about user behavior. Finally, the low-frequency data features are input into the pre-trained value assessment model to obtain a more accurate assessment result. The present application effectively alleviates the problem of insufficient coverage of low-frequency data, reduces assessment bias, and realizes accurate quantification of the value of low-frequency transaction data.
[0134] In this embodiment, the electronic device (eg, Figure 1 The server shown in the figure) can receive instructions or obtain data through a wired connection or a wireless connection. It should be noted that the above wireless connection method may include but is not limited to 3G / 4G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other wireless connection methods currently known or to be developed in the future.
[0135] It should be emphasized that in order to further ensure the privacy and security of the above transaction data information, the above transaction data information can also be stored in a blockchain node.
[0136] The blockchain referred to in this application is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.
[0137] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
[0138] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.
[0139] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).
[0140] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.
[0141] Further references Figure 3 , as a response to the above Figure 2 The present application provides an embodiment of a data value assessment device, and the device embodiment is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.
[0142] like Figure 3 As shown, the data value assessment device 300 described in this embodiment includes:
[0143] The data classification module 301 is used to obtain transaction data and transaction frequency influencing factors, and classify the transaction data according to the transaction frequency influencing factors to obtain high-frequency transaction data and low-frequency transaction data;
[0144] A high frequency analysis module 302 is used to perform feature analysis on high frequency trading data and obtain features of high frequency trading data;
[0145] A graph construction module 303 is used to construct a data relationship graph of high-frequency transaction data based on the high-frequency transaction data characteristics, wherein the nodes of the data relationship graph represent users or commodities, and the edges of the data relationship graph represent transaction relationships;
[0146] A graph traversal module 304 is used to perform a graph traversal operation on the data relationship graph using a graph neural network algorithm to obtain implicit association features between users and products;
[0147] A low-frequency analysis module 305 is used to perform feature analysis on low-frequency transaction data based on implicit correlation features using a transfer learning algorithm to obtain features of the low-frequency transaction data;
[0148] The value assessment module 306 is used to import the low-frequency transaction data features into the pre-trained data value assessment model to obtain the value assessment results of the low-frequency transaction data.
[0149] Furthermore, the data classification module 301 is specifically used for:
[0150] Acquire transaction data and clean the transaction data to obtain standardized transaction data; obtain matching transaction frequency influencing factors according to data classification requirements, and calculate the transaction frequency of each transaction data based on the transaction frequency influencing factors; compare the transaction frequency of each transaction data with the preset transaction frequency threshold to obtain high-frequency transaction data and low-frequency transaction data.
[0151] Furthermore, the graph construction module 303 is specifically used for:
[0152] Classify the high-frequency transaction data features to obtain user data features, commodity data features and transaction relationship features; initialize a graph object, and create graph nodes in the graph object based on the user data features and commodity data features; calculate the weights of the edges, and generate the edges of the data relationship graph based on the edge weights and transaction relationship features; embed the edges of the data relationship graph into the graph object to obtain the data relationship graph of the high-frequency transaction data.
[0153] Furthermore, the graph construction module 303 is further used to:
[0154] Identify transaction frequency and transaction amount in high-frequency transaction data; calculate the weight value of transaction frequency to obtain a first weight value; calculate the weight value of transaction amount to obtain a second weight value; sum the first weight value and the second weight value to obtain the weight of the edge; use the edge weight to weight the transaction relationship characteristics, and draw the edges of the data relationship map based on the weighted transaction relationship characteristics.
[0155] Furthermore, the graph traversal module 304 is specifically used for:
[0156] The format of the relationship graph is converted into a graph data format; the relationship graph converted into a graph data format is imported into a preset initial feature matrix to obtain a graph feature matrix; the graph feature matrix is input into a pre-trained graph neural network, and the convolution layer of the graph neural network is used to traverse and convolve the graph feature matrix to obtain implicit association features.
[0157] Furthermore, the graph neural network is calculated based on the graph neural network algorithm, and the graph neural network includes a first convolution layer and a second convolution layer, the first convolution layer is used to convolve the user data features, and the second convolution layer is used to convolve the product data features, and the graph traversal module 304 is further used to:
[0158] A preset sliding window is used to traverse the graph feature matrix to identify the user data feature submatrix and the product data feature submatrix; a first convolutional layer is used to perform a convolution operation on the user data feature submatrix to obtain a first convolutional feature; a second convolutional layer is used to perform a convolution operation on the product data feature submatrix to obtain a second convolutional feature; the first convolutional feature and the second convolutional feature are fused to obtain an implicit association feature.
[0159] Furthermore, the low frequency analysis module 305 is specifically used for:
[0160] Determine the data type label that matches the low-frequency transaction data and obtain the target data type label;
[0161] Determine the high-frequency trading data matching the target data type label, and obtain the target high-frequency trading data;
[0162] Obtaining implicit correlation features corresponding to target high-frequency trading data to obtain target implicit correlation features;
[0163] Extract features of low-frequency transaction data to obtain initial data features;
[0164] The initial data features and the target implicit correlation features are fused to obtain the frequent transaction data features.
[0165] In the above embodiment, the present application discloses a data value assessment device, which belongs to the field of artificial intelligence technology. The present application first classifies transaction data based on transaction frequency to separate high-frequency and low-frequency transaction data. Then, by constructing a data relationship graph for high-frequency transaction data and traversing the graph using a graph neural network algorithm, the implicit correlation features between users and commodities are extracted. The implicit features in these high-frequency transaction data reflect user behavior and market dynamics. Afterwards, through transfer learning, these implicit features are migrated to low-frequency transaction data, so that low-frequency data can have richer feature representations under limited coverage and capture deep information about user behavior. Finally, the low-frequency data features are input into the pre-trained value assessment model to obtain a more accurate assessment result. The present application effectively alleviates the problem of insufficient coverage of low-frequency data, reduces assessment bias, and realizes accurate quantification of the value of low-frequency transaction data.
[0166] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.
[0167] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 4 with a memory 41, a processor 42, and a network interface 43, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.
[0168] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.
[0169] The memory 41 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, 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 (FlashCard), etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit of the computer device 4 and its external storage device. In this embodiment, the memory 41 is generally used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions of the data value assessment method, etc. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or are to be output.
[0170] The processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run the computer-readable instructions stored in the memory 41 or process data, such as running the computer-readable instructions of the data value assessment method.
[0171] The network interface 43 may include a wireless network interface or a wired network interface. The network interface 43 is generally used to establish a communication connection between the computer device 4 and other electronic devices.
[0172] In the above embodiment, the present application discloses a computer device, which belongs to the field of artificial intelligence technology. The present application first classifies transaction data based on transaction frequency to separate high-frequency and low-frequency transaction data. Then, by constructing a data relationship graph for high-frequency transaction data and traversing the graph using a graph neural network algorithm, the implicit correlation features between users and commodities are extracted. The implicit features in these high-frequency transaction data reflect user behavior and market dynamics. Afterwards, through transfer learning, these implicit features are migrated to low-frequency transaction data, so that low-frequency data can have richer feature representations under limited coverage and capture deep information about user behavior. Finally, the low-frequency data features are input into a pre-trained value assessment model to obtain a more accurate assessment result. The present application effectively alleviates the problem of insufficient coverage of low-frequency data, reduces assessment bias, and realizes accurate quantification of the value of low-frequency transaction data.
[0173] The present application also provides another embodiment, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the data value assessment method as described above.
[0174] In the above embodiment, the present application discloses a computer-readable storage medium, which belongs to the field of artificial intelligence technology. The present application first classifies transaction data based on transaction frequency to separate high-frequency and low-frequency transaction data. Then, by constructing a data relationship graph for high-frequency transaction data and traversing the graph using a graph neural network algorithm, the implicit correlation features between users and commodities are extracted. The implicit features in these high-frequency transaction data reflect user behavior and market dynamics. Afterwards, through transfer learning, these implicit features are migrated to low-frequency transaction data, so that low-frequency data can have richer feature representations under limited coverage and capture deep information about user behavior. Finally, the low-frequency data features are input into a pre-trained value assessment model to obtain a more accurate assessment result. The present application effectively alleviates the problem of insufficient coverage of low-frequency data, reduces assessment bias, and realizes accurate quantification of the value of low-frequency transaction data.
[0175] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.
[0176] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.
[0177] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.
Claims
1. A data value assessment method, characterized in that: include: Acquiring transaction data and transaction frequency influencing factors, and classifying the transaction data according to the transaction frequency influencing factors to obtain high-frequency transaction data and low-frequency transaction data; Performing feature analysis on the high-frequency trading data to obtain features of the high-frequency trading data; Based on the high-frequency transaction data features, construct a data relationship graph of the high-frequency transaction data, wherein the nodes of the data relationship graph represent users or commodities, and the edges of the data relationship graph represent transaction relationships; Using a graph neural network algorithm to perform a graph traversal operation on the data relationship graph to obtain implicit association features between users and products; Based on the implicit correlation features, using a transfer learning algorithm to perform feature analysis on the low-frequency transaction data to obtain low-frequency transaction data features; The low-frequency transaction data features are imported into a pre-trained data value assessment model to obtain a value assessment result of the low-frequency transaction data.
2. The data value assessment method according to claim 1, characterized in that: The step of acquiring transaction data and transaction frequency influencing factors, and classifying the transaction data according to the transaction frequency influencing factors to obtain high-frequency transaction data and low-frequency transaction data specifically includes: Acquiring transaction data, and performing data cleaning on the transaction data to obtain standardized transaction data; According to the data classification requirements, a matching transaction frequency impact factor is obtained, and the transaction frequency of each of the transaction data is calculated based on the transaction frequency impact factor; The transaction frequency of each of the transaction data is compared with a preset transaction frequency threshold to obtain the high-frequency transaction data and the low-frequency transaction data.
3. The data value assessment method according to claim 1, characterized in that: The step of constructing a data relationship graph of the high-frequency trading data based on the high-frequency trading data features specifically includes: Classifying the high-frequency transaction data features to obtain user data features, commodity data features, and transaction relationship features; Initializing a graph object, and creating graph nodes in the graph object based on the user data features and the product data features; Calculating edge weights, and generating edges of the data relationship graph based on the edge weights and the transaction relationship characteristics; The edges of the data relationship graph are embedded into the graph object to obtain the data relationship graph of the high-frequency trading data.
4. The data value assessment method according to claim 3, characterized in that: The step of calculating the weight of the edge and generating the edge of the data relationship graph based on the weight of the edge and the transaction relationship feature specifically includes: Identifying the transaction frequency and transaction amount in the high-frequency trading data; Calculating the weight value of the transaction frequency to obtain a first weight value; Calculating a weight value of the transaction amount to obtain a second weight value; Performing weighted summation on the first weight value and the second weight value to obtain the weight of the edge; The transaction relationship features are weighted using the weights of the edges, and the edges of the data relationship graph are drawn based on the weighted transaction relationship features.
5. The data value assessment method according to claim 1, characterized in that: The step of using a graph neural network algorithm to perform a graph traversal operation on the data relationship graph to obtain implicit association features between users and products specifically includes: Converting the format of the relationship graph into a graph data format; Importing the relationship graph converted into a graph data format into a preset initial feature matrix to obtain a graph feature matrix; The graph feature matrix is input into a pre-trained graph neural network, and the convolution layer of the graph neural network is used to traverse and convolve the graph feature matrix to obtain the implicit association features.
6. The data value assessment method according to claim 5, characterized in that: The graph neural network is calculated based on the graph neural network algorithm, and the graph neural network includes a first convolution layer and a second convolution layer, the first convolution layer is used to convolve the user data features, and the second convolution layer is used to convolve the product data features. The step of using the convolution layer of the graph neural network to traverse and convolve the graph feature matrix to obtain the implicit association features specifically includes: Using a preset sliding window to traverse the graph feature matrix, identifying a user data feature submatrix and a product data feature submatrix; Using the first convolution layer to perform a convolution operation on the user data feature submatrix to obtain a first convolution feature; Using the second convolutional layer to perform a convolution operation on the commodity data feature submatrix to obtain a second convolutional feature; The first convolution feature and the second convolution feature are subjected to feature fusion to obtain the implicit association feature.
7. The data value assessment method according to claim 1, characterized in that: The step of performing feature analysis on the low-frequency transaction data using a transfer learning algorithm based on the implicit association feature to obtain the feature of the low-frequency transaction data specifically includes: Determine a data type label that matches the low-frequency transaction data to obtain a target data type label; Determine high-frequency trading data that matches the target data type label to obtain target high-frequency trading data; Obtaining the implicit correlation features corresponding to the target high-frequency trading data to obtain the target implicit correlation features; Extracting features of the low-frequency transaction data to obtain initial data features; The initial data features and the target implicit association features are subjected to feature fusion to obtain the frequency transaction data features.
8. A data value assessment device, characterized in that: include: A data classification module, used to obtain transaction data and transaction frequency influencing factors, and classify the transaction data according to the transaction frequency influencing factors to obtain high-frequency transaction data and low-frequency transaction data; A high-frequency analysis module, used to perform feature analysis on the high-frequency trading data to obtain features of the high-frequency trading data; A graph construction module, configured to construct a data relationship graph of the high-frequency transaction data based on the high-frequency transaction data features, wherein the nodes of the data relationship graph represent users or commodities, and the edges of the data relationship graph represent transaction relationships; A graph traversal module, used to perform a graph traversal operation on the data relationship graph using a graph neural network algorithm to obtain implicit association features between users and products; A low-frequency analysis module, used to perform feature analysis on the low-frequency transaction data using a transfer learning algorithm based on the implicit correlation features to obtain low-frequency transaction data features; The value assessment module is used to import the low-frequency transaction data features into a pre-trained data value assessment model to obtain the value assessment results of the low-frequency transaction data.
9. A computer device, characterized in that: It comprises a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the data value assessment method as described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the data value assessment method according to any one of claims 1 to 7.
