Data asset recommendation method, equipment and medium

Abstract

The invention provides a data asset recommendation method and device and a medium, and relates to the field of data processing. The method comprises the steps of generating at least one theme of data assets according to production side features and consumption side features of the data assets; determining a dynamic weight of the data assets according to the production side features and the consumption side features of the data assets, and generating first data asset recommendation information according to the dynamic weight of the data assets; generating second data asset recommendation information according to the information input by the user and a similarity algorithm; according to the historical consumption data of the user, third data asset recommendation information is generated, and the data asset recommendation information comprises the theme of the recommended data assets. According to the method, the comprehensiveness and accuracy of data asset recommendation are improved.

Description

Technical Field

[0001] This application relates to the field of data processing, and more particularly to a method, apparatus, and medium for recommending data assets. Background Technology

[0002] As enterprises advance their digital transformation, the scale and complexity of data assets continue to increase, making the efficient integration, management, and utilization of these assets a key challenge.

[0003] In related technologies, data asset recommendations typically rely solely on users' historical consumption behavior, assigning fixed weights to the tracked data of users' historical consumption behavior for recommendations. However, this method of data asset recommendation is neither comprehensive nor accurate.

[0004] Therefore, a recommendation scheme for data assets is needed that can improve the comprehensiveness and accuracy of recommendations. Summary of the Invention

[0005] This application provides a method, device, and medium for recommending data assets, which can improve the comprehensiveness and accuracy of data asset recommendations.

[0006] In a first aspect, embodiments of this application provide a method for recommending data assets, including:

[0007] Based on the production-side and consumption-side characteristics of the data asset, at least one theme of the data asset is generated;

[0008] The dynamic weight of the data asset is determined based on the production-side and consumption-side characteristics of the data asset, and the first data asset recommendation information is generated based on the dynamic weight of the data asset.

[0009] Based on user input and a similarity algorithm, generate second-generation data asset recommendation information;

[0010] Based on users' historical consumption data, third-party data asset recommendation information is generated, which includes the themes of the recommended data assets.

[0011] In one possible implementation, generating at least one theme of the data asset based on its production-side and consumption-side characteristics includes:

[0012] Feature extraction is performed on the production-side data of data assets to obtain the production-side features of data assets;

[0013] Feature extraction is performed on the consumption-side data of data assets to obtain the consumption-side features of data assets;

[0014] Based on the clustering algorithm that maximizes mutual information and the mutual information clustering loss function, the production-side features and consumption-side features of the data assets are clustered to obtain the probability that the data assets belong to each theme.

[0015] At least one topic of the data asset is generated based on the topic corresponding to a probability greater than a preset probability threshold.

[0016] In one possible implementation, determining the dynamic weight of the data asset based on its production-side and consumption-side characteristics includes:

[0017] Based on the production-side and consumption-side characteristics of the data assets, a composite evaluation index for the data assets is determined;

[0018] Based on the composite evaluation index of the data assets, an initial decision matrix is ​​constructed, and the initial decision matrix is ​​standardized to obtain a standardized matrix.

[0019] The entropy value corresponding to the data asset is determined based on the standardized matrix, and the initial weight of the data asset is determined based on the entropy value.

[0020] The initial weights are dynamically adjusted based on the time decay coefficient, user feedback coefficient, and asset click-through rate of the data assets to determine the dynamic weights of the data assets.

[0021] In one possible implementation, generating first data asset recommendation information based on the dynamic weights of the data assets includes:

[0022] The weighted normalization matrix is ​​determined based on the dynamic weights of the data assets;

[0023] The positive and negative ideal solutions are determined based on the weighted normalization matrix.

[0024] The proximity of the data asset is determined based on the positive and negative ideal solutions;

[0025] Based on the order of relevance of the data assets from high to low and the number of recommendation digits, first data asset recommendation information is generated.

[0026] In one possible implementation, generating second data asset recommendation information based on user input and a similarity algorithm includes:

[0027] Extract keywords from user-input information to obtain keyword information;

[0028] The similarity between the asset data and the keyword information is determined based on the cosine similarity algorithm;

[0029] Based on the order of similarity between the asset data and the keyword information from high to low, and the recommendation digit, second data asset recommendation information is generated.

[0030] In one possible implementation, generating third-party data asset recommendation information based on user historical consumption data includes:

[0031] Based on preset time thresholds and users' historical consumption data, short-term historical consumption data and long-term historical consumption data are determined.

[0032] The short-term historical consumption data is input into a time-aware Transformer encoder to obtain the first consumption data asset.

[0033] The long-term historical consumption data is input into a deep decomposition matrix enhanced by GNN to obtain the second consumption data asset;

[0034] Identify other data assets that overlap with, but are not entirely identical to, the themes of the first and second consumer data assets, and generate third data asset recommendation information based on these other data assets.

[0035] In one possible implementation, the production-side characteristics include one or more of the following: data quality, timeliness, metadata semantics, lineage, and update frequency;

[0036] The consumer-side features include one or more of the following: usage scenarios, usage frequency, data analysis, operation fields, authoritativeness of producer-side contributors, and update frequency.

[0037] Secondly, embodiments of this application provide a computing device, including:

[0038] The topic generation module is used to generate at least one topic of the data asset based on the production-side characteristics and consumption-side characteristics of the data asset.

[0039] The data recommendation module is used to determine the dynamic weight of the data asset based on its production-side and consumption-side characteristics, and generate first data asset recommendation information based on the dynamic weight of the data asset; generate second data asset recommendation information based on user input information and a similarity algorithm; and generate third data asset recommendation information based on the user's historical consumption data. The data asset recommendation information includes the topic of the recommended data asset.

[0040] Thirdly, embodiments of this application provide another computing device, including:

[0041] The processor, and the memory that is in communication with the processor;

[0042] Memory is used to store instructions that the computer executes;

[0043] The processor is configured to execute computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0044] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect described above.

[0045] Fifthly, embodiments of this application provide a computer program product, including a computer program, which, when executed by a processor, is used to implement the first aspect and / or various possible implementations of the first aspect as described above.

[0046] This application provides a method, device, and medium for recommending data assets. First, based on the production-side and consumption-side characteristics of the data assets, at least one theme for the data assets is generated. This allows the theme to comprehensively consider information from the entire production and consumption chain, achieving joint modeling of technological value and consumer popularity value, thus improving the comprehensiveness and accuracy of data asset recommendations. Subsequently, different data asset recommendation information can be generated based on the dynamic weights of the data assets, user input information, and user historical consumption data. This diversified recommendation method better supports personalized user choices, further improving the comprehensiveness and accuracy of data asset recommendations and increasing the success rate of asset recommendations. Furthermore, recommending based on the dynamic weights of the data assets can adapt to dynamic changes in the data assets, further improving the accuracy of data asset recommendations. Attached Figure Description

[0047] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0048] Figure 1 This is a system architecture diagram of an embodiment of this application;

[0049] Figure 2 A flowchart illustrating a data asset recommendation method according to an embodiment of this application;

[0050] Figure 3 This is a schematic diagram illustrating the data asset recommendation process according to an embodiment of this application;

[0051] Figure 4 This is a schematic diagram of the structure of a computing device according to an embodiment of this application;

[0052] Figure 5 This is a schematic diagram of the structure of a computing device according to another embodiment of this application.

[0053] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0054] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0055] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0056] The collection, storage, use, processing, transmission, provision, and disclosure of financial data or user data involved in the technical solution of this application all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0057] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, they do not mean that the applicant has used or necessarily used the solution.

[0058] It should also be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with relevant laws, regulations and standards, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0059] The data asset recommendation method, equipment, and media of this application can be used in the field of data processing, or in any field other than data processing, such as the field of information recommendation. The application field of the data asset recommendation method, equipment, and media of this application is not limited.

[0060] The data asset recommendation method, equipment, and medium of this application can be applied to scenarios where banks and other financial institutions recommend data assets. The data asset recommendation method, equipment, and medium of this application can be applied to any scenario involving data asset recommendation.

[0061] First, let me explain the terms used in this application:

[0062] Data assets: Data assets are formed after data is generated and then designed and processed.

[0063] Production side: The producers of data assets.

[0064] Consumer side: The consumers of data assets, such as downstream consumption systems and data analysts.

[0065] Asset Themes: Data assets are divided into themes based on data source, processing logic, and application, such as loan themes and merchant themes.

[0066] As enterprises advance their digital transformation, the scale and complexity of data assets continue to increase, making the efficient integration, management, and utilization of these assets a key challenge. Production-side data assets are typically generated by data engineers and analysts, while the consumption-side involves the data needs of data consumers (such as business departments and decision-makers).

[0067] Traditional data asset management systems often face problems such as data silos and low data quality, making it difficult to meet the needs of cross-departmental collaboration and data-driven decision-making. In related technologies, data asset recommendations typically rely solely on users' historical consumption behavior, assigning fixed weights to the tracked data of users' historical consumption behavior for recommendations.

[0068] However, this method of recommending data assets ignores the production characteristics and technical characteristics of data assets themselves, such as data quality, timeliness, and authority, resulting in a lack of technical value in the recommendation results and low comprehensiveness and accuracy of data asset recommendations.

[0069] Based on the aforementioned technical problems, the inventive concept of this application is to provide a recommendation scheme for data assets that can improve the comprehensiveness and accuracy of recommendations.

[0070] This application provides a method, device, and medium for recommending data assets. First, based on the production-side and consumption-side characteristics of the data assets, at least one theme for the data assets is generated. This allows the theme to comprehensively consider information from the entire production and consumption chain, achieving joint modeling of technological value and consumer popularity value, thus improving the comprehensiveness and accuracy of data asset recommendations. Subsequently, different data asset recommendation information can be generated based on the dynamic weights of the data assets, user input information, and user historical consumption data. This diversified recommendation method better supports personalized user choices, further improving the comprehensiveness and accuracy of data asset recommendations and increasing the success rate of asset recommendations. Furthermore, recommending based on the dynamic weights of the data assets can adapt to dynamic changes in the data assets, further improving the accuracy of data asset recommendations.

[0071] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0072] Figure 1 This is a system architecture diagram of one embodiment of this application, such as... Figure 1 As shown, when data asset recommendations need to be made to users, the terminal device sends a data asset recommendation instruction to the server. Upon receiving the instruction, the server generates at least one topic for the data asset based on its production and consumption characteristics. It then determines the dynamic weight of the data asset based on these characteristics and generates first data asset recommendation information. Next, it generates second data asset recommendation information based on user input and a similarity algorithm. Finally, it generates third data asset recommendation information based on the user's historical consumption data. The data asset recommendation information includes the topic of the recommended data asset. The server sends the first, second, and third data asset recommendation information to the terminal device, which displays the data asset recommendation information output by each recommendation method on different display interfaces.

[0073] Figure 2 This is a flowchart illustrating a data asset recommendation method according to an embodiment of this application. This embodiment describes the data asset recommendation method using a computing device as the executing entity. Figure 2 As shown, the recommendation method for this data asset may include the following steps:

[0074] S201: Generate at least one theme of the data asset based on the production-side and consumption-side characteristics of the data asset.

[0075] In this embodiment, the computing device can be a terminal device such as a computer or tablet, or a server-side device such as a cloud server or physical server, or an integrated terminal-server device. As long as the computing device can perform distributed database data processing, no restrictions are imposed here.

[0076] In this embodiment, the production-side features can be technical features, reflecting the technical aspects of production, while the consumption-side features can be behavioral features, reflecting the behavior of consumers.

[0077] In this embodiment, the production-side features can be obtained by extracting features from the production-side information, such as extracting keywords from the production-side information, or by extracting features using the DataRank method, which is an improvement on PageRank.

[0078] In this embodiment, the consumer-side features can be obtained by feature extraction from consumer-side information, such as keyword extraction from consumer-side information, or extraction based on the HHEM (Hierarchical Hotness Evaluation Model).

[0079] In this embodiment, a single data asset may belong to one or more themes.

[0080] S202: Determine the dynamic weight of the data asset based on the production-side and consumption-side characteristics of the data asset, and generate the first data asset recommendation information based on the dynamic weight of the data asset.

[0081] In this embodiment, the data asset recommendation information includes the topic of the recommended data assets.

[0082] In this embodiment, multiple evaluation indicators corresponding to the data assets can be determined based on the production-side and consumption-side characteristics of the data assets, and an original matrix can be constructed accordingly. The fixed weights of the data assets can be calculated based on the original matrix and the entropy-weight method. By introducing time decay coefficients, user feedback coefficients, etc., the fixed weights can be dynamically adjusted to obtain the dynamic weights of the data assets.

[0083] In this embodiment, fixed weights cannot adapt to the dynamic changes of data assets. For example, newly launched data needs to focus on technical value, while declining data needs to have its popularity weight reduced. Therefore, the weights of data assets can be dynamically adjusted to improve the accuracy of recommendations based on weights.

[0084] S203: Generate second data asset recommendation information based on user input and similarity algorithm.

[0085] In this embodiment, similarity calculations can be performed on the information input by the user and various data assets to recommend data assets similar to the information input by the user, thereby achieving personalized data asset recommendations.

[0086] S204: Generate third-party data asset recommendation information based on users' historical consumption data.

[0087] In this embodiment, recommendations can be made based on users' historical consumption data, enabling recommendations to be made based on preferences of consumer-side data assets.

[0088] In this embodiment, after comprehensively considering the entire supply chain information from both the production and consumption sides to determine the subject of data asset ownership, data assets can be recommended based on three different methods. This diversified recommendation approach better supports personalized user choices. The computing device can display the data asset recommendation information output by each method on different display interfaces, allowing users to click on the corresponding interface to view the relevant recommendations based on their needs.

[0089] For example, Figure 3 This is a schematic diagram of a data asset recommendation process according to an embodiment of this application, such as... Figure 3 As shown, the computing device can first perform A. Intelligent topic allocation of data assets to determine the topic to which the data assets belong, and then perform B. Integrated topic asset recommendation to obtain the first data asset recommendation information, C. Content-based asset recommendation to obtain the second data asset recommendation information, and D. Asset recommendation based on historical consumption to obtain the third data asset recommendation information.

[0090] In this embodiment, at least one theme for the data asset can first be generated based on its production-side and consumption-side characteristics. This allows the theme to comprehensively consider information from the entire production and consumption chain, enabling joint modeling of technological value and consumer popularity value, thus improving the comprehensiveness and accuracy of data asset recommendations. Subsequently, different data asset recommendation information can be generated based on the dynamic weights of the data asset, user input information, and the user's historical consumption data. This diversified recommendation method better supports personalized user choices, further improving the comprehensiveness and accuracy of data asset recommendations and increasing the success rate of asset recommendations. Furthermore, recommending based on the dynamic weights of the data asset can adapt to dynamic changes in the data asset, further improving the accuracy of data asset recommendations.

[0091] In one possible implementation, production-side characteristics may include one or more of the following: data quality, timeliness, metadata semantics, lineage, and update frequency.

[0092] Consumer-side characteristics may include one or more of the following: usage scenarios, usage frequency, data analysis, operation fields, authoritativeness of producer-side contributors, and update frequency.

[0093] In this embodiment, the production-side features can be technical features that reflect the technical nature of the production side. They are not limited to data quality, timeliness, metadata semantics, lineage, update frequency, etc. Those skilled in the art can flexibly set production-side features according to actual needs, and no restrictions are imposed here.

[0094] In this embodiment, consumer-side characteristics can be behavioral characteristics that reflect consumer behavior. They are not limited to usage scenarios, usage frequency, data analysis, operation fields, the authority of producer-side contributors, update frequency, etc. Those skilled in the art can flexibly set consumer-side characteristics according to actual needs, and no restrictions are imposed here.

[0095] In this embodiment, production-side features can be technical features that reflect the technical aspects of production, such as data quality, timeliness, metadata semantics, lineage, and update frequency. Consumer-side data can be behavioral features that reflect consumer-side behavior, such as usage scenarios, usage frequency, data analysis, operation fields, authority of production-side contributors, and update frequency.

[0096] In one possible implementation, step S201, which generates at least one topic of the data asset based on the production-side and consumption-side characteristics of the data asset, may include:

[0097] S11: Extract features from the production-side data of data assets to obtain the production-side features of data assets.

[0098] S12: Extract features from the consumption-side data of data assets to obtain the consumption-side features of data assets.

[0099] S13: Based on the clustering algorithm that maximizes mutual information and the mutual information clustering loss function, cluster the production-side and consumption-side features of data assets to obtain the probability of data assets belonging to each theme.

[0100] S14: Generate at least one topic of the data asset based on the topic corresponding to a probability greater than a preset probability threshold.

[0101] In this embodiment, the production-side features can be obtained by extracting features from the production-side information, such as extracting keywords from the production-side information, or by extracting features using the DataRank method improved from PageRank.

[0102] In this embodiment, the consumer-side features can be obtained by feature extraction from consumer-side information, such as keyword extraction from consumer-side information, or extraction based on the HHEM model (Hierarchical Hotness Evaluation Model).

[0103] In this embodiment, the MIMC (Mutual Information Maximization based Clustering) algorithm can be used to cluster the production-side and consumption-side characteristics of data assets to obtain the probability of data assets belonging to each theme.

[0104] In this embodiment, the mutual information clustering loss function can be expressed by the following formula (1):

[0105] L_cluster = a×I(h_p; h_c) + b×[I(h_p); h_p'] + c×I(h_c; h_c') (1)

[0106] Where L_cluster represents the loss value, I(h_p; h_c) represents the mutual information between the production side and the consumption side, [I(h_p); h_p'] represents the stability of the production side features, I(h_c; h_c') represents the stability of the consumption side features, and a, b, and c represent the weight parameters.

[0107] In this embodiment, the preset probability threshold can be flexibly set by those skilled in the art according to actual conditions. For example, the preset probability threshold can be 0.7 or 0.8, and no restrictions are imposed here.

[0108] For example, if the output of a clustering algorithm for a certain data asset is [Topic A: 0.82], [Topic B: 0.73], [Topic C: 0.65], [Topic D: 0.57], ..., and the preset probability threshold is 0.7, then the data asset belongs to the themes A and B.

[0109] In this embodiment, dual-stream feature extraction can be performed on the production-side and consumption-side information of data assets to obtain technical features reflecting the technical aspects of production and behavioral features reflecting the behavior of consumption. Then, clustering is performed using a mutual information maximization clustering algorithm and a mutual information clustering loss function to obtain the probability of each data asset belonging to a specific theme. By filtering these probabilities according to a preset probability threshold, the theme to which the data asset belongs can be determined.

[0110] In one possible implementation, the step S202 above, which involves determining the dynamic weight of the data asset based on its production-side and consumption-side characteristics, may include:

[0111] S21: Determine composite evaluation indicators for data assets based on the production-side and consumption-side characteristics of data assets.

[0112] S22: Based on the composite evaluation index of data assets, construct an initial decision matrix and standardize the initial decision matrix to obtain a standardized matrix.

[0113] S23: Determine the entropy value corresponding to the data asset based on the standardized matrix, and determine the initial weight of the data asset based on the entropy value.

[0114] S24: Based on the time decay coefficient of data assets, user feedback coefficient, and asset click-through rate, dynamically adjust the initial weights to determine the dynamic weights of data assets.

[0115] In this embodiment, the process of determining the dynamic weights of data assets may include:

[0116] (1) Based on the production-side and consumption-side characteristics of data assets and the following formula (2), determine the composite evaluation index for each data asset:

[0117] r ij =fx ij1 +gx ij2 (2)

[0118] (2) Construct an initial decision matrix based on the composite evaluation index of each data asset, and standardize the initial decision matrix to obtain a standardized matrix.

[0119] (3) The entropy value corresponding to the data asset is calculated using the following formulas (3)-(4):

[0120]

[0121] (4) Calculate the initial weights of the data assets based on the entropy value and the following formula (5):

[0122]

[0123] (5) Introducing the time decay coefficient, user feedback coefficient, and asset click-through rate, the dynamic weight of the data asset is calculated using the following formula (6):

[0124]

[0125] Where γ represents the time decay coefficient, β represents the user feedback coefficient, and click-through rate represents the asset click-through rate.

[0126] In this embodiment, the time decay coefficient and user feedback coefficient can be flexibly set by those skilled in the art according to actual conditions. For example, the decay of consumer-side indicators in the time decay coefficient can be faster than that of production-side indicators, and the user feedback coefficient can be dynamically optimized based on A / B testing.

[0127] In this embodiment, fixed weights cannot adapt to the dynamic changes of data assets. For example, newly launched data needs to focus on technical value, while declining data needs to have its popularity weight reduced. Therefore, after calculating the initial weight of the data asset using the above formulas (2)-(5), time decay coefficient, user feedback coefficient, asset click rate, etc. can be introduced to dynamically adjust the initial weight, so that the dynamic weight adapts to the dynamic changes of the data asset, improves the accuracy of the dynamic weight, and thus improves the accuracy of data asset recommendation using dynamic weight.

[0128] In one possible implementation, the step S202 above, generating the first data asset recommendation information based on the dynamic weights of the data assets, may include:

[0129] S31: Determine the weighted normalization matrix based on the dynamic weights of the data assets.

[0130] S32: Determine the positive and negative ideal solutions based on the weighted normalization matrix.

[0131] S33: Determine the proximity of data assets based on the positive and negative ideal solutions.

[0132] S34: Generate the first data asset recommendation information based on the order of data asset relevance from high to low and the recommendation number.

[0133] In this embodiment, the proximity of the data asset can be calculated using the following formulas (7)-(11):

[0134] (1) Determine the weighted normalization matrix using the following formula (7) and dynamic weights:

[0135]

[0136] (2) Use the following formula (8) to determine the positive ideal solution and the negative ideal solution:

[0137]

[0138] (3) The proximity of the data assets is calculated using the following formulas (9)-(11):

[0139]

[0140] In this embodiment, a "cold start, hot cycle" integrated theme asset recommendation method is proposed. Cold start refers to calculating a preliminary asset recommendation benchmark based on the technical characteristics of the production side when a new data asset has not yet been consumed. Hot cycle refers to determining dynamic weights based on user behavior data features such as consumption frequency and correlation analysis, and continuously updating and iterating the asset recommendation order and ranking.

[0141] In this implementation, after determining the dynamic weights, the weighted normalization matrix, positive ideal solution, and negative ideal solution can be determined based on the dynamic weights, thereby determining the proximity of each data asset. Then, a comprehensive TOPSIS evaluation is performed on each data asset. By using the order of proximity from high to low and the recommendation number, the data assets to be recommended can be determined.

[0142] In one possible implementation, step S203, which generates second data asset recommendation information based on user input and a similarity algorithm, may include:

[0143] S41: Extract keywords from the user input information to obtain keyword information.

[0144] S42: Determine the similarity between asset data and keyword information based on the cosine similarity algorithm.

[0145] S43: Generate second-level asset recommendation information based on the order of similarity between asset data and keyword information from high to low, and the recommendation digit.

[0146] In this embodiment, a cosine similarity algorithm based on TF-IDF can be used to calculate the similarity between keyword information and data asset metadata.

[0147] In this embodiment, the recommended number of digits can be flexibly set by those skilled in the art according to actual needs. For example, it can be the first 20 digits or the first 10 digits, and no restrictions are imposed here.

[0148] In this embodiment, similarity calculations can be performed on the information input by the user and various data assets. Recommendations are made based on the semantic similarity between the asset data and the user, from high to low. Data assets similar to the information input by the user are recommended, thereby achieving personalized data asset recommendations and further improving the comprehensiveness and accuracy of data asset recommendations.

[0149] In one possible implementation, step S204 above, which generates third-party data asset recommendation information based on the user's historical consumption data, may include:

[0150] S51: Determine short-term historical consumption data and long-term historical consumption data based on preset time thresholds and user historical consumption data.

[0151] S52: Input short-term historical consumption data into a time-aware Transformer encoder to obtain the first consumption data asset.

[0152] S53: Input long-term historical consumption data into a deep decomposition matrix based on GNN enhancement to obtain the second consumption data asset.

[0153] S54: Identify other data assets that overlap with, but are not entirely the same as, the themes of the first and second consumer data assets, and generate third data asset recommendation information based on these other data assets.

[0154] In this embodiment, the preset time threshold can be flexibly set by those skilled in the art according to actual conditions. For example, the preset time threshold can be 3 days or 7 days, and no restrictions are imposed here.

[0155] In this embodiment, if the theme corresponding to the first consumer data asset is A and B, and the theme of the second consumer data asset is C and D, then other data assets are assets that include at least one of A / B / C / D, and are not A, B, C, or D. That is, they overlap with but are not completely the same as the themes of the first and second consumer data assets.

[0156] In this implementation, short-term historical consumption data can be analyzed using a time-aware Transformer encoder to obtain the user's first short-term consumption data asset, and long-term historical consumption data can be analyzed using a deep decomposition matrix based on GNN enhancement to obtain the user's second long-term consumption data asset. Based on other data assets that overlap with but are not completely the same as the first and second consumption data assets, recommendations can be made based on the preferences of consumption-side data assets, thereby further improving the comprehensiveness and accuracy of data asset recommendations.

[0157] The following describes the data asset recommendation method of this application using a specific embodiment.

[0158] In a specific embodiment, the recommendation process for a bank's data assets is as follows:

[0159] The first step involves the computing device extracting features from the production-side data of the data assets to obtain the production-side features of the data assets; and extracting features from the consumption-side data of the data assets to obtain the consumption-side features of the data assets.

[0160] The second step involves the computing device clustering the production-side and consumption-side features of the data assets based on a clustering algorithm that maximizes mutual information and a mutual information clustering loss function, thereby obtaining the probability that the data assets belong to each theme; and generating at least one theme for the data assets based on the themes corresponding to probabilities greater than a preset probability threshold.

[0161] The third step involves the computing device determining composite evaluation indicators for data assets based on their production and consumption characteristics. Based on these composite evaluation indicators, an initial decision matrix is ​​constructed and standardized to obtain a standardized matrix. The entropy value corresponding to the data asset is determined based on the standardized matrix, and the initial weight of the data asset is determined based on the entropy value. Finally, the initial weight is dynamically adjusted based on the data asset's time decay coefficient, user feedback coefficient, and click-through rate to determine the dynamic weight of the data asset.

[0162] The fourth step involves the computing device determining a weighted normalization matrix based on the dynamic weights of the data assets; determining positive and negative ideal solutions based on the weighted normalization matrix; determining the proximity of the data assets based on the positive and negative ideal solutions; and generating the first data asset recommendation information based on the order of proximity of the data assets from high to low and the number of recommendation bits.

[0163] The fifth step involves the computing device extracting keywords from the user-input information to obtain keyword information; determining the similarity between asset data and keyword information using a cosine similarity algorithm; and generating second-level asset recommendation information based on the order of similarity between asset data and keyword information from high to low and the recommendation digit.

[0164] The sixth step involves the computing device determining short-term and long-term historical consumption data based on preset time thresholds and user historical consumption data. The short-term historical consumption data is then input into a time-aware Transformer encoder to obtain the first consumption data asset. The long-term historical consumption data is input into a deep decomposition matrix enhanced by GNN to obtain the second consumption data asset. Other data assets that overlap with but are not completely identical to the themes of the first and second consumption data assets are identified, and third data asset recommendation information is generated based on these other data assets.

[0165] The seventh step involves the computing device displaying the data asset recommendation information output by each recommendation method on different display interfaces.

[0166] Figure 4 This is a schematic diagram of the structure of a computing device according to an embodiment of this application, as shown below. Figure 4 As shown, the computing device includes: a topic generation module 41, used to generate at least one topic of the data asset based on the production-side characteristics and consumption-side characteristics of the data asset; a data recommendation module 42, used to determine the dynamic weight of the data asset based on the production-side characteristics and consumption-side characteristics of the data asset, and generate first data asset recommendation information based on the dynamic weight of the data asset; generate second data asset recommendation information based on user input information and a similarity algorithm; and generate third data asset recommendation information based on the user's historical consumption data, wherein the data asset recommendation information includes the topic of the recommended data asset.

[0167] The computing device provided in this application embodiment can execute the technical solution shown in the above method embodiment. Its implementation principle and beneficial effects are similar, and will not be repeated here.

[0168] Figure 5 This is a schematic diagram of the structure of a computing device according to another embodiment of this application, such as... Figure 5 As shown, the computing device includes: a processor 501 and a memory 502 communicatively connected to the processor 501; the memory 502 stores computer execution instructions; the processor 501 executes the computer execution instructions stored in the memory 502 to implement the steps of the data asset recommendation method in the above method embodiments.

[0169] In the aforementioned computing device, the memory 502 and the processor 501 are electrically connected directly or indirectly to enable data transmission or interaction. For example, these components can be electrically connected to each other via one or more communication buses or signal lines, such as a bus connection. The memory 502 stores computer execution instructions that implement data access control methods, including at least one software functional module that can be stored in the memory 502 in the form of software or firmware. The processor 501 executes various functional applications and data processing by running the software programs and modules stored in the memory 502.

[0170] The memory 502 may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc. The memory 502 stores programs, which are then executed by the processor 501 upon receiving execution instructions. Furthermore, the software programs and modules within the memory 502 may include an operating system, which may include various software components and / or drivers for managing system tasks (e.g., memory management, storage device control, power management, etc.) and can communicate with various hardware or software components to provide an operating environment for other software components.

[0171] Processor 501 can be an integrated circuit chip with signal processing capabilities. The aforementioned processor 501 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor.

[0172] An embodiment of this application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the steps of the various method embodiments of this application.

[0173] An embodiment of this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the various method embodiments of this application.

[0174] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0175] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0176] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.

[0177] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or as software program modules.

[0178] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0179] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the appended claims.

[0180] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for recommending data assets, characterized in that, include: Based on the production-side and consumption-side characteristics of the data asset, at least one theme of the data asset is generated; The dynamic weight of the data asset is determined based on the production-side and consumption-side characteristics of the data asset, and the first data asset recommendation information is generated based on the dynamic weight of the data asset. Based on user input and a similarity algorithm, generate second-generation data asset recommendation information; Based on users' historical consumption data, third-party data asset recommendation information is generated, which includes the themes of the recommended data assets.

2. The data asset recommendation method according to claim 1, characterized in that, The step of generating at least one theme for the data asset based on its production-side and consumption-side characteristics includes: Feature extraction is performed on the production-side data of data assets to obtain the production-side features of data assets; Feature extraction is performed on the consumption-side data of data assets to obtain the consumption-side features of data assets; Based on the clustering algorithm that maximizes mutual information and the mutual information clustering loss function, the production-side features and consumption-side features of the data assets are clustered to obtain the probability that the data assets belong to each theme. At least one topic of the data asset is generated based on the topic corresponding to a probability greater than a preset probability threshold.

3. The data asset recommendation method according to claim 1, characterized in that, The step of determining the dynamic weight of the data asset based on its production-side and consumption-side characteristics includes: Based on the production-side and consumption-side characteristics of the data assets, a composite evaluation index for the data assets is determined; Based on the composite evaluation index of the data assets, an initial decision matrix is ​​constructed, and the initial decision matrix is ​​standardized to obtain a standardized matrix. The entropy value corresponding to the data asset is determined based on the standardized matrix, and the initial weight of the data asset is determined based on the entropy value. The initial weights are dynamically adjusted based on the time decay coefficient, user feedback coefficient, and asset click-through rate of the data assets to determine the dynamic weights of the data assets.

4. The data asset recommendation method according to claim 3, characterized in that, The step of generating first data asset recommendation information based on the dynamic weights of the data assets includes: The weighted normalization matrix is ​​determined based on the dynamic weights of the data assets; The positive and negative ideal solutions are determined based on the weighted normalization matrix. The proximity of the data asset is determined based on the positive and negative ideal solutions; Based on the order of relevance of the data assets from high to low and the number of recommendation digits, first data asset recommendation information is generated.

5. The method for recommending data assets according to any one of claims 1-4, characterized in that, The step of generating second data asset recommendation information based on user input and a similarity algorithm includes: Extract keywords from user-input information to obtain keyword information; The similarity between the asset data and the keyword information is determined based on the cosine similarity algorithm; Based on the order of similarity between the asset data and the keyword information from high to low, and the recommendation digit, second data asset recommendation information is generated.

6. The method for recommending data assets according to any one of claims 1-4, characterized in that, The process of generating third-party data asset recommendation information based on users' historical consumption data includes: Based on preset time thresholds and users' historical consumption data, short-term historical consumption data and long-term historical consumption data are determined. The short-term historical consumption data is input into a time-aware Transformer encoder to obtain the first consumption data asset. The long-term historical consumption data is input into a deep decomposition matrix enhanced by GNN to obtain the second consumption data asset; Identify other data assets that overlap with, but are not entirely the same as, the themes of the first and second consumer data assets, and generate third data asset recommendation information based on these other data assets.

7. The method for recommending data assets according to any one of claims 1-4, characterized in that, The production-side characteristics include one or more of the following: data quality, timeliness, metadata semantics, lineage, and update frequency; The consumer-side features include one or more of the following: usage scenarios, usage frequency, data analysis, operation fields, authoritativeness of producer-side contributors, and update frequency.

8. A computing device, characterized in that, include: The topic generation module is used to generate at least one topic of the data asset based on the production-side characteristics and consumption-side characteristics of the data asset. The data recommendation module is used to determine the dynamic weight of the data asset based on the production-side and consumption-side characteristics of the data asset, and generate first data asset recommendation information based on the dynamic weight of the data asset; and generate second data asset recommendation information based on user input information and a similarity algorithm. Based on users' historical consumption data, third-party data asset recommendation information is generated, which includes the themes of the recommended data assets.

9. A computing device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory is used to store computer-executed instructions; The processor is configured to execute computer execution instructions stored in the memory, causing the processor to perform the recommended method for data assets as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the recommended method for the data asset as described in any one of claims 1-7.

Images