Database table classification treatment method and system based on multi-modal fusion

Abstract

The invention relates to the technical field of electric power data management, and provides a database table classification management method and system based on multi-modal fusion. The method comprises the following steps: extracting table information in a table in data, including metadata information and historical sql information of a table; on the basis of the extracted metadata information and historical sql information, constructing a feature vector of multi-modal fusion; for the existing database tables in the initial data table, based on the feature vectors of the database tables, calculating the distance between the feature vectors of each table, and adopting a hierarchical clustering method to perform clustering analysis on the data tables; specifying a storage scheme for each type of data table in the clustering result, and forming a storage scheme result of the existing database table; constructing an attention-based classification model based on a storage scheme result of the existing database table; and classifying the tables newly added into the database of the data table by adopting the classification model, and determining a storage scheme result of the newly added tables. According to the invention, the most reasonable storage scheme can be automatically allocated to the table.

Description

Technical Field

[0001] The present invention relates to the technical field of power data management, and in particular to a database table classification management method and system based on multimodal fusion. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] With the rapid development of information technology, the application of data middle platform in the power industry is becoming more and more extensive, and it has gradually become a key factor in improving the competitiveness of enterprises. As a comprehensive platform for centralized management and utilization of data resources, the data middle platform not only greatly enhances the ability of power companies to process complex data, but also significantly improves their data analysis level and decision-making efficiency, providing a solid foundation for the intelligent transformation of power companies. However, with the deepening of digital transformation, the amount of data generated by the power industry has shown an explosive growth trend, which has posed unprecedented challenges to data management. The traditional data management method based on a single relational database has been unable to cope with the needs of massive data storage and real-time analysis faced by modern data middle platforms due to its limitations in scalability, processing speed and cost-effectiveness. Therefore, exploring and practicing more efficient data management technologies and architectures has become a core issue that needs to be urgently solved in the current construction of data middle platforms in the power industry.

[0004] Since there are many types of business data in the power industry, including user information, electricity consumption data, equipment information, etc., these data have different data scales and access requirements. Therefore, the use of multi-type storage systems can store the data in different storage devices according to its characteristics to improve data access efficiency and management effects. For example, data with small data scale and frequent changes (such as user information data and equipment information data) can be stored in traditional relational databases such as Oracle, and data with large data scale, less changes and low access requirements (such as historical electricity consumption data) can be stored in data based on distributed file storage such as Hive.

[0005] Specifically, for information with small data size but frequent updates, such as user personal information and device status information, traditional relational database systems (such as Oracle) are most suitable. Such systems are known for their powerful transaction processing capabilities, which can ensure data consistency and integrity while providing efficient query performance, making them very suitable for processing small-scale data sets that require frequent reading and writing. For data with large data volumes, few updates, and relatively low access requirements, such as historical electricity consumption data, data warehouse technology based on distributed file systems (such as Hive) is more suitable. Such systems are designed with large-scale data storage and analysis requirements in mind, and can provide excellent scalability and cost-effective data storage solutions. By storing these historical data in a distributed environment, efficient data management and in-depth analysis can be achieved, thereby providing valuable insights for power companies to help them optimize resource allocation, predict future trends, and make more scientific decisions. In addition, with the development of IoT technology, the power industry is also facing more and more unstructured data processing needs, such as video surveillance data, sensor data, etc. For such data, more flexible storage solutions such as NoSQL databases or object storage can be considered to meet their special data structures and access patterns.

[0006] However, in actual applications, due to the large size of data tables in the database, it is not feasible to select a suitable storage solution for each table through manual analysis. Summary of the invention

[0007] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a database table classification management method and system based on multimodal fusion. The present invention can automatically allocate the most reasonable storage solution to the table, help database administrators to realize intelligent data classification management, and improve the efficiency of data access and analysis.

[0008] In order to achieve the above object, the present invention adopts the following technical solution:

[0009] The first aspect of the present invention provides a database table classification management method based on multimodal fusion.

[0010] A database table classification management method based on multimodal fusion, comprising:

[0011] Extract table information from the data center, including table metadata and historical SQL information;

[0012] Construct a multimodal fusion feature vector based on the extracted metadata information and historical SQL information;

[0013] For the existing database tables in the initial data, based on their feature vectors, the distance between the feature vectors of each table is calculated, and the hierarchical clustering method is used to perform cluster analysis on the data tables;

[0014] For each type of data table in the clustering result, a storage plan is specified for it to form a storage plan result for the existing database table;

[0015] Based on the storage solution results of the existing database tables, an attention-based classification model is constructed; the classification model is used to classify the tables in the database newly added to the data center to determine the storage solution results of the newly added tables.

[0016] Furthermore, the metadata information of the table includes: table name, table description information, table field information, table size and table foreign keys; the historical SQL information of the table includes: the number of add, delete, modify and query statements of the table and table information of associated queries.

[0017] Furthermore, the feature vectors represent different types of information through numerical values ​​and sequences, respectively, wherein:

[0018] The unique identifier of the table, the size of the table, the proportion of select statements in historical SQL information, the proportion of insert statements, and the proportion of update statements and delete statements are numeric;

[0019] The foreign key of a table is represented by a sequence of variable length, and the value in the sequence is the table ID with dependency relationship.

[0020] Table comments and field names are represented by a coded sequence of variable length, where each value in the sequence is numeric;

[0021] The table information of the associated query is represented by a sequence of variable length, and each value in the sequence is a key-value pair.

[0022] Furthermore, the distance between the feature vectors of each table is calculated based on its feature vector, and the method includes:

[0023] For numerical eigenvectors, subtraction is used to calculate the distance of the vector in the dimension;

[0024] For non-numeric eigenvectors, a specific processing method is used to calculate the distance of the vector in that dimension;

[0025] The specific processing method is used to calculate the distance of the vector in this dimension, and the method includes: for the dependency characteristics of the table, if the two tables have a dependency relationship, its value is set to 0, otherwise it is set to 1; for the table comments and field names, based on the shortest sequence, the closest values ​​are selected in the long sequence to form a sequence of equal length, and then the cosine similarity is calculated; for the table information of the associated query, if there is an associated query between the two tables, the number of queries is used as the distance value, otherwise it is set to 0.

[0026] Furthermore, the hierarchical clustering method is used to perform cluster analysis on the data table, and the method includes:

[0027] Treat each table as a class, calculate the distance between each class using the weighted distance formula, and store it;

[0028] Merge the two classes with the smallest distance into one class, and calculate the distance between the merged new class and the remaining classes based on the average method;

[0029] Repeat the previous process until only the required n classes remain.

[0030] Furthermore, the classification model is constructed through an attention-based recurrent neural network.

[0031] The second aspect of the present invention provides a database table classification management system based on multimodal fusion.

[0032] A database table classification management system based on multimodal fusion, including:

[0033] The data information extraction module is configured to: extract table information in the data center, including metadata information and historical SQL information of the table;

[0034] A feature vector construction module is configured to: construct a multimodal fusion feature vector based on the extracted metadata information and historical SQL information;

[0035] The hierarchical clustering module is configured to: for the existing database tables in the initial data, based on their feature vectors, calculate the distance between the feature vectors of each table, and perform cluster analysis on the data table using the hierarchical clustering method;

[0036] The classification module is configured to: specify a storage scheme for each type of data table in the clustering result, and form a storage scheme result of an existing database table;

[0037] A multi-type storage module is configured to: build an attention-based classification model based on the storage solution results of existing database tables; use the classification model to classify the tables in the database newly added to the data center, and determine the storage solution results of the newly added tables.

[0038] A third aspect of the present invention provides a computer-readable storage medium.

[0039] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps in the database table classification management method based on multimodal fusion as described in the first aspect above.

[0040] A fourth aspect of the present invention provides a computer device.

[0041] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the database table classification management method based on multimodal fusion as described in the first aspect above are implemented.

[0042] A fifth aspect of the present invention provides a computer program product or a computer program.

[0043] The present invention provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps in the database table classification management method based on multimodal fusion as described in the first aspect above.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] First, traditional database management methods often rely on manual operations, which is not only time-consuming and labor-intensive, but also prone to errors due to human factors. The introduction of intelligent classification governance technology has greatly reduced the need for manual intervention through automation and intelligent means. For example, intelligent algorithms can automatically identify and classify data, reduce the storage of duplicate data, and promptly detect and correct erroneous records. This not only reduces the workload of managers, but also significantly improves the accuracy and efficiency of data management.

[0046] Secondly, intelligent classification governance technology can automatically select the most suitable storage media and database type according to the different characteristics and usage frequency of the data. For example, for frequently accessed hot data, you can choose a high-performance memory database or solid-state drive storage; while for cold data with low access frequency, you can use lower-cost tapes or cloud storage. This flexible data storage strategy not only saves storage costs, but also improves data access speed and system response time.

[0047] Finally, through intelligent classification management, the system can dynamically adjust the data index and cache strategy according to the user's query mode and data access habits. For example, commonly used data can be pre-loaded into the cache to reduce disk I / O operations; while complex query requests can quickly locate the required data through the optimized index structure. This not only improves the speed of data access, but also enhances the overall performance of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0049] Figure 1 It is an overall flow chart of a database table classification management method based on multimodal fusion provided by an embodiment of the present invention;

[0050] Figure 2 It is a structural diagram of a database table classification management system based on multimodal fusion provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0052] It should be noted that the following detailed descriptions are all illustrative and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0053] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0054] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the methods and systems according to various embodiments of the present disclosure. It should be noted that each box in the flowchart or block diagram can represent a module, a program segment, or a part of a code, and the module, program segment, or a part of a code may include one or more executable instructions for implementing the logical functions specified in each embodiment. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, or they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the flowchart and / or block diagram, and the combination of boxes in the flowchart and / or block diagram can be implemented using a dedicated hardware-based system that performs a specified function or operation, or can be implemented using a combination of dedicated hardware and computer instructions.

[0055] Embodiment 1

[0056] like Figure 1 As shown, this embodiment provides a database table classification management method based on multimodal fusion. This embodiment uses the method applied to the server as an example. It can be understood that the method can also be applied to terminals, and can also be applied to terminals, servers and systems, and implemented through the interaction between terminals and servers. The server can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network servers, cloud communications, middleware services, domain name services, security services CDN, and big data and artificial intelligence platforms. The terminal can be a smart phone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, etc., but is not limited to this. The terminal and the server can be directly or indirectly connected via wired or wireless communication, which is not limited in this application. In this embodiment, the method includes the following steps:

[0057] Step 1: Data table information extraction: obtain metadata information and historical SQL information of the Taichung data table in the data;

[0058] For each table, its metadata information includes table name, table comment, field name, table size, and foreign key information; its historical SQL information includes: number of queries, number of additions, number of deletions and modifications, and table information of associated queries, including the table name and number of queries of associated queries.

[0059] Step 2: Feature vector construction: construct a multimodal fusion feature vector based on the extracted metadata information and historical SQL information;

[0060] For the table names and field names in the table metadata information and historical SQL information, they are encoded by building a unified ID. For other text information, they are encoded by word segmentation and dictionary construction.

[0061] So for each table T i , which can be expressed as:

[0062] T i ={i,ZS i ,ZD i ,K i ,s i ,sl i ,in i ,ud i ,R i}

[0063] Among them, i,s i Respectively represent the unique identifier of the table and the size of the table; sl i ,in i ,ud i Respectively represent the proportion of select statements, insert statements, update statements and delete statements in historical SQL; ZS i and ZD i Respectively represent the encoding information of table comments and field names, which are sequences of variable length; K i Representation and Table T i Table information with dependencies such as foreign keys, which is also a sequence of variable length; R i Representation and Table T i The table information for associated query is also a sequence of variable length, but each value in the sequence is a key-value pair of <table ID, query count>.

[0064] Step 3: Hierarchical clustering: Based on its eigenvector, calculate the distance between the eigenvectors of each table, and use the hierarchical clustering method to complete the clustering analysis of the data table. Specifically, the hierarchical clustering method includes the following steps:

[0065] Step 301: First, treat each table feature vector as a class, calculate the distance between each class, and store it. When calculating the distance between each feature vector, the distance between feature vectors is calculated using a weighted distance formula. In particular, for numerical feature vectors in the feature vector, the distance of the vector in this dimension is directly calculated by subtraction; for non-numerical feature vectors, the distance of the vector in this dimension is first calculated using a specific processing method, and the specific processing method includes:

[0066] Regarding the dependency characteristics of tables, if two tables T iand T j There is a dependency relationship, whose value is D_K i Set to 0, otherwise set to 1;

[0067] For features such as table annotations and field names, since their length is not fixed, the shortest sequence is used as the benchmark, and the closest value is selected from the long sequence to form a sequence of equal length, and then the cosine similarity is calculated, and this value is taken as the distance between feature values ​​such as table annotations and field names, D_ZD ij ;

[0068] For the table information of the associated query, if there is an associated query between two tables, the distance value D_R ij The number of common queries for the two tables, otherwise set to 0.

[0069] In summary, two classes (that is, two tables) T i and T j The distance calculation formula is as follows:

[0070]

[0071] Among them, α 1 ~α 8 are weight coefficients for different features, which can be manually trained and adjusted on small-scale samples and finally applied to global clustering.

[0072] Step 302: The two classes T with the smallest distance i and T j Aggregate into a new class T ij , and recalculate the distance between the merged new class and the remaining classes. The average distance method is used for calculation, that is, the aggregated class T ij With other types of T * The distance formula is as follows:

[0073]

[0074] Among them, D i* and D j* Respectively represent the class T before aggregation i and T j With class T * distance.

[0075] Step 303: Repeat step 302 until only the required n classes remain.

[0076] Step 4: Determine the storage plan. For the existing database tables in the data center, the database administrator analyzes the common characteristics of each type of data table in the clustering results, and specifies a storage plan for it based on its characteristics to form the storage plan results for the existing database tables. For the database tables newly incorporated into the data center, a storage plan for the newly incorporated database tables is formed based on the classification results and the storage plan that has been determined for the classification.

[0077] Step 5: Build a classification model. Using the storage scheme after hierarchical clustering as sample data, a classification model is built through an attention-based recurrent neural network.

[0078] Step 6: Classification: Based on the classification model, classify the newly incorporated database table to determine which category in the clustering results the newly incorporated database table belongs to.

[0079] The present invention takes the electric power information collection system database as an example, and stores the data originally stored only in the Oracle database through the Oracle and Hive databases. Based on the database table classification management method based on multimodal fusion proposed by the present invention, the optimal storage solution for the table is determined. When the total server resources consumed by the two storage solutions are equal, the SQL execution performance comparison is as follows:

[0080] Table 2 Comparison of SQL execution performance after data classification and governance

[0081] Original storage solution New storage solution Change Statement 0.087 0.091 Query Statement 0.991 0.86

[0082] According to the results in Table 2, based on the database table classification management method based on multimodal fusion proposed by the present invention, the performance of change statements is basically not affected when stored through a new multi-type storage solution, and the performance of query statements is improved by about 13% compared with the original storage solution.

[0083] Embodiment 2

[0084] like Figure 2 As shown, this embodiment provides a database table classification management system based on multimodal fusion.

[0085] A database table classification management system based on multimodal fusion, including:

[0086] The data information extraction module is configured to: extract table information in the data center, including metadata information and historical SQL information of the table;

[0087] A feature vector construction module is configured to: construct a multimodal fusion feature vector based on the extracted metadata information and historical SQL information;

[0088] The hierarchical clustering module is configured to: for the existing database tables in the initial data, based on their feature vectors, calculate the distance between the feature vectors of each table, and perform cluster analysis on the data table using the hierarchical clustering method;

[0089] The classification module is configured to: specify a storage scheme for each type of data table in the clustering result, and form a storage scheme result of an existing database table;

[0090] A multi-type storage module is configured to: build an attention-based classification model based on the storage solution results of existing database tables; use the classification model to classify the tables in the database newly added to the data center, and determine the storage solution results of the newly added tables.

[0091] In some embodiments, the metadata information of the table includes: table name, table description information, table field information, table size and table foreign keys; the historical SQL information of the table includes: the number of add, delete, modify and query statements of the table and table information of associated queries.

[0092] In some embodiments, the feature vectors represent different types of information through numerical values ​​and sequences, respectively, wherein:

[0093] The unique identifier of the table, the size of the table, the proportion of select statements in historical SQL information, the proportion of insert statements, and the proportion of update statements and delete statements are numeric;

[0094] The foreign key of a table is represented by a sequence of variable length, and the value in the sequence is the table ID with dependency relationship.

[0095] Table comments and field names are represented by a coded sequence of variable length, where each value in the sequence is numeric;

[0096] The table information of the associated query is represented by a sequence of variable length, and each value in the sequence is a key-value pair.

[0097] In some embodiments, the hierarchical clustering module is further configured to: for a numerical feature vector, use subtraction to calculate the distance of the vector in a dimension; for a non-numerical feature vector, use a specific processing method to calculate the distance of the vector in the dimension; wherein the specific processing method is used to calculate the distance of the vector in the dimension, and the method includes: for the dependency feature of the table, if there is a dependency between the two tables, its value is set to 0, otherwise it is set to 1; for table comments and field names, take the shortest sequence as the basis, select the closest numerical value in the long sequence to form a sequence of equal length, and then calculate the cosine similarity; for the table information of the associated query, if there is an associated query between the two tables, the number of queries is used as the distance value, otherwise it is set to 0.

[0098] In some embodiments, the hierarchical clustering module is further configured to: regard each table as a class, calculate the distance between each class by a weighted distance formula, and store it; merge the two classes with the smallest distance into one class, and calculate the distance between the merged new class and the remaining classes based on the average method; repeat the previous step until only the required n classes remain. Further, the classification model is constructed by an attention-based recurrent neural network.

[0099] Embodiment 3

[0100] This embodiment provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the database table classification management method based on multimodal fusion as described in the above-mentioned embodiment 1 are implemented.

[0101] Embodiment 4

[0102] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps in the database table classification management method based on multimodal fusion as described in the first embodiment above are implemented.

[0103] Embodiment 5

[0104] This embodiment provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the steps in the above-mentioned database table classification management method based on multimodal fusion.

[0105] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of hardware embodiments, software embodiments, or embodiments combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage, etc.) containing computer-usable program code.

[0106] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0107] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0108] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0109] A person 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 a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0110] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A database table classification management method based on multimodal fusion, characterized in that: include: Extract table information from the data center, including table metadata and historical SQL information; Construct a multimodal fusion feature vector based on the extracted metadata information and historical SQL information; For the existing database tables in the initial data, based on their feature vectors, the distance between the feature vectors of each table is calculated, and the hierarchical clustering method is used to perform cluster analysis on the data tables; For each type of data table in the clustering result, a storage plan is specified for it to form a storage plan result for the existing database table; Based on the storage solution results of the existing database tables, an attention-based classification model is constructed; the classification model is used to classify the tables in the database newly added to the data center to determine the storage solution results of the newly added tables.

2. The database table classification management method based on multimodal fusion according to claim 1 is characterized in that: The metadata information of the table includes: table name, table description information, table field information, table size and table foreign keys; the historical SQL information of the table includes: the number of add, delete, modify and query statements of the table and table information of associated queries.

3. The database table classification management method based on multimodal fusion according to claim 1 is characterized in that: The feature vectors represent different types of information through numerical values ​​and sequences, respectively, where: The unique identifier of the table, the size of the table, the proportion of select statements in historical SQL information, the proportion of insert statements, and the proportion of update statements and delete statements are numeric; The foreign key of a table is represented by a sequence of variable length, and the value in the sequence is the table ID with dependency relationship. Table comments and field names are represented by a coded sequence of variable length, where each value in the sequence is numeric; The table information of the associated query is represented by a sequence of variable length, and each value in the sequence is a key-value pair.

4. The database table classification management method based on multimodal fusion according to claim 3 is characterized in that: The method of calculating the distance between the feature vectors of each table based on the feature vectors thereof includes: For numerical eigenvectors, subtraction is used to calculate the distance of the vector in the dimension; For non-numeric eigenvectors, a specific processing method is used to calculate the distance of the vector in that dimension; The specific processing method is used to calculate the distance of the vector in this dimension, and the method includes: for the dependency characteristics of the table, if the two tables have a dependency relationship, its value is set to 0, otherwise it is set to 1; for the table comments and field names, based on the shortest sequence, the closest values ​​are selected in the long sequence to form a sequence of equal length, and then the cosine similarity is calculated; for the table information of the associated query, if there is an associated query between the two tables, the number of queries is used as the distance value, otherwise it is set to 0.

5. The database table classification management method based on multimodal fusion according to claim 1 is characterized in that: The hierarchical clustering method is used to perform cluster analysis on the data table, and the method includes: Treat each table as a class, calculate the distance between each class using the weighted distance formula, and store it; Merge the two classes with the smallest distance into one class, and calculate the distance between the merged new class and the remaining classes based on the average method; Repeat the previous process until only the required n classes remain.

6. The database table classification management method based on multimodal fusion according to claim 1 is characterized in that: The classification model is constructed through an attention-based recurrent neural network.

7. A database table classification management system based on multimodal fusion, characterized in that: include: The data information extraction module is configured to: extract table information in the data center, including metadata information and historical SQL information of the table; A feature vector construction module is configured to: construct a multimodal fusion feature vector based on the extracted metadata information and historical SQL information; The hierarchical clustering module is configured to: for the existing database tables in the initial data, based on their feature vectors, calculate the distance between the feature vectors of each table, and perform cluster analysis on the data table using the hierarchical clustering method; The classification module is configured to: specify a storage scheme for each type of data table in the clustering result, and form a storage scheme result of an existing database table; A multi-type storage module is configured to: build an attention-based classification model based on the storage solution results of existing database tables; use the classification model to classify the tables in the database newly added to the data center, and determine the storage solution results of the newly added tables.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the database table classification management method based on multimodal fusion as described in any one of claims 1 to 6 are implemented.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the database table classification management method based on multimodal fusion as described in any one of claims 1-6 are implemented.

10. A computer program product, characterized in that The computer program product includes a computer program, and when the computer program is executed by a processor, the steps in the database table classification management method based on multimodal fusion as described in any one of claims 1 to 6 are implemented.

Images