Index calculation processing engine and method based on index wide table blurring data model

Abstract

The invention discloses an index calculation processing engine and method based on an index wide table blurring data model, and relates to the technical field of index middle tables. The engine comprises a business index definition module, a cue word driving module, a large language model module, an index semantic data creation module, an index semantic model generation module and an index calculation unit module. According to the method, the pain points of a current system platform in the aspects of index processing, data response, operation complexity, calculation efficiency and the like are effectively solved flexibly through intelligent, automatic and semantic technical means. A more flexible, efficient and intelligent index management application solution is provided for enterprises, and the enterprises are helped to keep competitive power in quickly changing business environments and grasp key business opportunities.

Description

Technical Field

[0001] The present invention belongs to the technical field of indicator middle platform, and in particular, relates to an indicator calculation and processing engine based on an indicator wide table virtualized data model and an implementation method based on the indicator wide table virtualized data model. Background Art

[0002] At present, the indicator platform is playing an increasingly critical role as an important platform for the digital transformation of enterprises. The existing indicator platform has not only made many efforts and improvements in the centralized management of enterprise indicators, but also realized the following core functions in the application of indicator calculation and processing engine: (1) Data aggregation: The data aggregation function is the cornerstone of the indicator calculation engine, which enables the indicator platform to perform fast aggregation operations on large amounts of data. For example, an enterprise may need to calculate the total sales of a quarter or the average transaction amount of a month. Aggregation operations include not only basic summation and averaging, but also statistical functions such as median and mode; (2) Data filtering: The data filtering function enables business departments to filter data according to specific business rules or query requirements. This is based on filtering for a specific date range, a specific category of products, or business indicator records that meet specific conditions; (3) Data sorting: The data sorting function allows business departments to sort data sets according to one or more fields, which helps to identify patterns or outliers in the data. For example, sorting customers by sales can help a company identify its most important customer groups, or optimizing inventory ABC management by sorting inventory items in descending order; (4) Complex calculations: Complex calculation functions support business departments to define and perform more complex mathematical or statistical operations, such as calculating growth rates, profit margins, or compound annual growth rates. These calculations usually require multi-step data processing and reveal deep indicators of business performance; (5) Multidimensional analysis: Multidimensional analysis allows business departments to view data from multiple perspectives, providing deep insights into the data. By slicing, dicing, and rotating on different dimensions, business departments can obtain richer pivot table views. For example, sales data can be analyzed by multiple dimensions such as time, region, product, and customer, thereby revealing the interrelationships and impacts between different factors; (6) Data visualization: Data visualization functions convert complex data into intuitive visual representations in the form of charts, graphs, and dashboards. This enables business departments to quickly identify key indicators and trends, facilitating more effective decision-making processes.

[0003] Through the empowerment of the above functions in the indicator center, the level of scientific decision-making support for enterprises has been improved, the data-driven cultural awareness within the enterprise has been enhanced, and the satisfaction of enterprise customers and risk management capabilities have been improved. By centrally managing and calculating key business indicators of enterprises, it can help enterprises monitor and analyze the current operating conditions of enterprises to a certain extent, and respond to market changes more quickly.

[0004] However, through the specific implementation and in-depth practical application of the indicator middle-office management platform, the defects and shortcomings of the current indicator middle-office application in the indicator calculation and processing engine are increasingly exposed: the indicator calculation and processing data model is rigid, cannot adapt to rapidly changing business needs, and lacks flexibility; when the business scenario undergoes major changes, it is usually necessary to redefine or generate the wide tables and summary tables that the indicator processing and calculation rely on; in complex scenarios, it is even necessary to rely on ETL engineers to reprocess the data that the indicators rely on. The entire processing process includes importing business system data, log data, and third-party data into the data warehouse to form new dimension tables and fact tables, and then generating new wide tables and summary tables based on these dimension tables and fact tables.

[0005] With the development of the indicator middle platform, the existing indicator management platform has increasingly exposed the following problems and pain points in the processing and calculation of indicators: The rigidity of the indicator data model is an increasingly obvious defect and problem faced by the existing indicator calculation and processing engine. In a rapidly changing business environment, enterprises need to adapt to changes in market dynamics and business needs in an agile manner. However, traditional indicator data models are often pre-defined, and this fixedness limits the ability of enterprises to respond quickly to data. When business needs change, traditional indicator data models may not be adjusted in time to reflect the new business logic, resulting in the inability of business indicator data analysis results to timely and accurately reflect the actual situation; as far as the current indicator management platform is concerned, when the business scenario changes significantly, it is usually necessary to redefine or generate the wide tables and summary tables that the indicator processing and calculation depends on. In complex scenarios, it is even necessary to rely on ETL engineers to reprocess the data that the indicators depend on. The entire processing process includes importing business system data, log data, and third-party data into the data warehouse to form new dimension tables and fact tables, and then generating new wide tables and summary tables based on these dimension tables and fact tables.

[0006] This inflexible way of working not only slows down the company's response to market changes, but also may miss key business opportunities. For example, when a company needs to quickly analyze new customer segments or evaluate the initial performance of a new product, a rigid indicator data model may not provide the necessary support. In addition, updating the data model in the indicator calculation engine usually requires the intervention of professionals, which not only increases the workload, but also increases the consumption of time and cost.

[0007] At the same time, in some cases, due to the lack of sufficient intelligence and flexibility, even simple indicator updates may require the intervention of professionals, which not only invisibly affects the business department's real-time demand for data, but also increases the difficulty of using the entire platform.

[0008] Therefore, based on the above background, the present invention invents an indicator calculation and processing engine and method based on an indicator wide table virtualization data model, thereby effectively solving the defects of the current indicator management platform in indicator calculation and processing. Summary of the invention

[0009] The present invention provides an indicator calculation and processing engine and method based on the indicator wide-table virtualized data model, which is used to solve the problem of indicator data model rigidity by intelligently and flexibly utilizing the indicator wide-table virtualized data model technology, thereby improving the application efficiency of the indicator calculation and processing engine in the overall indicator management platform and improving the usage experience of business departments.

[0010] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0011] The indicator calculation and processing engine based on the indicator wide table virtualization data model of the present invention includes:

[0012] Business indicator definition module: responsible for defining and creating new business indicators. This module allows business departments to set or optimize business indicators based on the specific business development needs of the enterprise, ensuring the accuracy and applicability of the indicators;

[0013] Prompt word driving module: It is used to generate keywords or phrases driven by business indicators based on the definition of business indicators, and to guide the creation of subsequent virtual wide tables and calculation expressions of indicators. Through the assembled prompt words, key information related to business indicators and indicator semantic model information are used to provide input for the large language model to generate virtual wide table structures and indicator calculation expressions of indicators.

[0014] Large language model module: uses natural language processing technology to conduct in-depth analysis and understanding of input prompt words; it can identify and process complex language structures, thereby accurately capturing the semantic information of business indicators, indicator semantics, and enterprise-related data elements, which not only provides support for the generation of indicator semantic models, but also ensures that the platform can accurately reflect the calculation of business indicators;

[0015] Indicator semantic data creation module: Based on the large language model module, it is responsible for creating and maintaining indicator semantic data; the indicator semantic data is the basis for enterprise data elements and enterprise indicator definitions, including detailed descriptions of daily business indicators, calculation methods and related parameters; with these data, the platform can more accurately calculate indicators within the enterprise;

[0016] Indicator semantic model generation module: Generates a specific indicator semantic model based on the created indicator semantic data, and expresses the model data in vectorized form; the indicator semantic model not only includes the definition of enterprise-related data elements, but also includes the definition of business indicators and the logic and rules required to calculate these indicators;

[0017] Indicator calculation unit module: responsible for performing actual indicator calculations, generating virtual wide table structures and indicator calculation expressions through indicator semantic models, and performing calculations and generating results based on actual business data;

[0018] Platform management users create indicator semantic data modules, integrate various enterprise data elements, and use the intelligence of the large language model module to further assist in the semantic understanding and generation of enterprise data and indicators. Finally, the indicator semantic model generation module stores all this information and calculation logic in the indicator semantic model knowledge base in the form of structured and vectorized data expressions, pre-building a rich enterprise indicator knowledge base for the entire platform, making the calculation of a wider range of business indicators more intelligent and flexible.

[0019] The platform business users start by defining new business indicators using the business indicator definition module. Driven by the business prompt word module, the semantics of the new business indicators are understood through the large language model, and the relevant enterprise indicator semantic knowledge is recalled. On this basis, a new virtualized wide table structure and indicator calculation expression for the business indicator calculation service are generated. The indicator calculation unit module performs specific calculation tasks, and on the basis of the virtualized wide table structure and indicator calculation expression, the corresponding enterprise data elements are integrated to generate the final business indicator calculation results.

[0020] Furthermore, the computer hardware system of the index calculation and processing engine includes a GPU server cluster and a K8S server cluster;

[0021] The large language model module runs on a GPU server cluster;

[0022] The business indicator definition module, prompt word driving module, indicator semantic model generation module, indicator semantic data creation module and indicator calculation unit module are respectively run on the K8S server cluster, and the definition of business indicators, the driving of prompt words, the generation of indicator semantic models, the definition of indicator semantic data and the implementation of indicator calculation are completed through Java and corresponding Springboot development components, and the communication between modules is completed by calling the API interface provided by each module;

[0023] The large language model module and the indicator semantic generation module complete the accurate capture and understanding of the semantic information of the business indicators involved, the indicator semantics and the enterprise-related data element information, and generate a specific indicator semantic model, expressing the model data in a vectorized form, ensuring the accuracy and consistency of the enterprise indicator calculation.

[0024] Furthermore, a file server cluster is also running on the K8S server cluster; the file server cluster is used to store virtualized wide table structures and indicator calculation expressions, and then provide file services for the indicator calculation unit module.

[0025] Furthermore, an enterprise data warehouse cluster is also running on the K8S server cluster, and the enterprise data warehouse cluster is used to store various enterprise data elements; if the enterprise data warehouse cluster already exists in the enterprise, it is directly connected to the indicator calculation and processing engine in the form of data consumption.

[0026] Furthermore, a management database cluster is also running on the K8S server cluster; the management database cluster stores structured data related to the platform of the present invention in a database cluster manner.

[0027] Furthermore, the management database cluster uses a MySQL cluster server as support.

[0028] The implementation method based on the indicator wide table virtual data model is implemented by using an indicator calculation processing engine based on the indicator wide table virtual data model, and includes the following steps:

[0029] S1. Create an indicator semantic model knowledge base through a large language model and collect business data from various business systems of the enterprise;

[0030] S2. The collected business data is processed to form enterprise business fact data and related dimension data;

[0031] S3. Use commonly used fact data and dimension data to generate wide tables and summary tables required for calculating the company's regular indicators;

[0032] S4. If the business indicator required by the business department is a regular indicator and can be directly calculated through regular wide tables and summary tables, it is directly calculated through the indicator calculation processing unit and the indicator calculation result is output; if the business indicator required by the business department cannot be directly calculated through regular wide tables and summary tables, a new business indicator needs to be created through the business department;

[0033] S5. Submit the newly created business indicator to the big language model. Since the big language model analyzes and creates the data elements required to calculate the business indicator, and combines the definition of the indicator semantic model knowledge base, prompts the vocabulary and other data elements of the enterprise, it dynamically creates a virtualized wide table structure and the business indicator calculation expression.

[0034] S6. Based on the virtualized wide table structure and the business indicator calculation expression, a new business indicator result is calculated and output through a business indicator calculation unit.

[0035] Furthermore, the implementation method is also divided into platform management user steps and platform business user steps:

[0036] Steps for platform business users:

[0037] M1. The platform management user starts the entire indicator semantic model knowledge base creation process: the platform management user integrates various enterprise data elements, creates and forms the indicator semantic data model;

[0038] M2, using the intelligence of large language models to further assist in the semantic understanding and generation of indicators;

[0039] M3. All this information and calculation logic are integrated into the indicator semantic model knowledge base in the form of vectorized data expression, providing a rich knowledge base for the entire platform, making the calculation of a wide range of business indicators more intelligent and flexible;

[0040] Steps for platform business users:

[0041] P1. Platform business users start by defining new business indicators. Driven by the prompt word driving module, they use the large language model to understand the semantics of new business indicators, recall relevant indicator semantic knowledge, and generate the virtualized wide table structure and indicator calculation expression of the business indicator calculation service;

[0042] P2, the indicator calculation unit module is responsible for executing specific calculation tasks. Based on the virtualized wide table structure and indicator calculation expression, it integrates enterprise data elements and generates the final enterprise indicator settlement results;

[0043] The virtualized wide table structure refers to a virtual wide table data structure dynamically generated according to the definition of business indicators; the indicator calculation expression is a specific business indicator calculation expression, which is generated corresponding to the virtualized wide table.

[0044] Furthermore, the indicator semantic data model is used to describe and define a structured model of data indicators, helping organizations to clearly express the calculation logic, data sources and business meanings of business indicators; the indicator semantic data model includes:

[0045] Business definition: a description of the business logic of the indicator, explaining the business meaning of the indicator and how it reflects business performance;

[0046] Calculation logic: the calculation formula or algorithm of the indicator, including the mathematical expression and logic used to calculate the indicator;

[0047] Data source: the source of raw data required for indicator calculation, including database tables, data warehouses or other data storage;

[0048] Dimensions and measures: dimensions and measures involved in the calculation of indicators. Dimensions include time, location, product category, etc.; measures include sales and profits, etc.;

[0049] Filter conditions: conditions used to limit the scope of metric calculation, including specific time ranges, geographic locations, or business departments;

[0050] Update frequency: The update frequency of indicator data, including daily, monthly or real-time updates.

[0051] Furthermore, the enterprise data elements refer to all business-related data of the enterprise, which are key components required by the enterprise when conducting business indicator management and analysis, covering the entire process from the generation, storage, processing to analysis and application of indicator data:

[0052] The enterprise data elements include:

[0053] Enterprise metadata: includes business metadata, technical metadata and operational metadata; business metadata is used to describe the business meaning and purpose of data, including business terms, definitions and rules of data; technical metadata is used to describe the technical details of data, including data model, database structure, data storage location and data conversion rules; operational metadata involves the management and maintenance process of data, including data creation, update, access and maintenance records;

[0054] Enterprise data types: including structured data, semi-structured data and unstructured data; the structured data is data stored in a relational database with a fixed format and pattern, including customer information tables and transaction records, etc.; the semi-structured data is data that does not fully conform to the relational database model, including XML and JSON documents, which have a structure but are not as strict as structured data; the unstructured data does not have a fixed format, including text files, pictures, videos and audio.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] (1) The present invention effectively responds to the changes in business indicators caused by the rapid changes in the business environment by introducing a dynamic and flexible indicator data model. In traditional indicator management platforms, the fixed and rigid data model leads to limited responsiveness to changes in enterprise market dynamics and business indicator requirements. The present invention allows instant adjustment and dynamic update of the indicator data model, allowing enterprises to quickly adapt to a wider range of business logic and ensure the timeliness and accuracy of business indicator analysis results. In the present invention, the indicator data model no longer uses a fixed or relatively fixed data model like the existing technology implementation method. The database table structure and calculation table that all business indicator calculations rely on in the existing technology are The expression is relatively fixed; in the present invention, the indicator data model will dynamically construct a virtualized wide table and indicator calculation expression based on the enterprise indicator semantic model and the new enterprise indicator definition, based on the natural language intent recognition and generation capabilities of the enterprise large language model, and submit it to the indicator calculation engine to generate results; on the basis of the existing platform, the present invention adds a business indicator definition module, a prompt word driving module, a large language model module, a module for creating indicator semantic data, an indicator semantic model generation module, and an indicator calculation unit module, thereby dynamically generating a virtualized wide table structure and an indicator calculation expression according to business indicator requirements; the definition and calculation of business indicators are no longer tightly coupled and dependent on pre-defined database tables;

[0057] (2) The present invention simplifies and reconstructs the indicator processing flow, reduces the reliance on professionals, and improves the intelligence and automation level of the flexible definition and calculation of business indicators. Under the existing platform, facing major changes in business scenarios, enterprises usually need to rely on ETL engineers to reprocess data or define data models. This process is not only time-consuming and labor-intensive, but may also miss key business opportunities. The present invention uses a dynamic and intelligent approach to implement the indicator data model. On the basis of various data elements of the enterprise, combined with the pre-generated enterprise indicator semantic model library of the present invention, the natural language intent recognition and generation capabilities of the enterprise's big language model are used to automatically calculate business indicators. In this process, many traditional tasks that rely on professionals are replaced by the intelligence of the enterprise's big language model. Therefore, while accelerating the enterprise's response to market changes, it also greatly improves the generalization processing capabilities for enterprise business indicators.

[0058] (3) The present invention enhances the ability to understand and express the business logic of an enterprise by pre-building an enterprise indicator semantic data model; this model can not only describe business needs more accurately, but also provide reusable business indicator analysis perspectives and elements; the indicator semantic data model of the present invention is constructed on the basis of various data elements of the enterprise. Therefore, compared with traditional wide tables and summary tables, the enterprise indicator semantic data model can more flexibly adapt to complex business scenarios and help enterprises better understand market dynamics and business trends;

[0059] (4) The present invention also improves the intelligence level of the platform by introducing a large language model and natural language processing technology, which enables non-professionals to easily define and update indicators, lowering the threshold for using the platform and also improving the satisfaction of business departments' real-time data needs; this intelligent data processing method not only reduces the workload of professionals, but also reduces the difficulty of using the entire platform;

[0060] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0062] Figure 1 It is a diagram of the fact data and dimension data structure of a certain deposit business in a specific embodiment of the present invention;

[0063] Figure 2 It is a virtual wide table structure and indicator calculation expression content diagram for the indicator of the amount of US dollar deposits of a business manager of a certain institution in the current month in a specific embodiment of the present invention;

[0064] Figure 3 It is a flowchart of the steps of the implementation method of the present invention based on the indicator wide table virtualization data model;

[0065] Figure 4 It is a general scheme diagram of the implementation method of the present invention based on the indicator wide table virtualization data model;

[0066] Figure 5 A working step diagram of the calculation program corresponding to the implementation method;

[0067] Figure 6 It is a program module structure diagram of the indicator calculation and processing engine based on the indicator wide table virtualization data model of the present invention;

[0068] Figure 7 The hardware structure diagram corresponding to the indicator calculation processing engine;

[0069] Figure 8 These are relatively fixed fact tables and dimension tables existing in the specific embodiment. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] See also Figure 1-8 As shown, the core problem solved by the present invention is: how to intelligently and flexibly utilize the indicator wide table virtualized data model technology to solve the problem of rigid indicator data model, thereby improving the application efficiency of the indicator calculation and processing engine in the overall indicator management platform and improving the usage experience of business departments.

[0072] Back to the business application scenarios of the indicator management platform, the increasingly exposed problem is that when the market changes, the problem that business personnel may face is that the original business indicators may lack or be insufficient to explain the current status of business development, so as to accurately and timely support business decisions. Therefore, business personnel need to develop new business indicators or optimize existing business indicators to accurately track the status of business development and changes. As a result, new demands for business indicator calculations are inevitable. If such new demands are calculated and processed using traditional indicator data ETL calculation and processing methods, the efficiency bottleneck is obvious.

[0073] Focusing on the above key scenarios, the present invention uses the following implementation to solve the deficiencies of the indicator calculation and processing engine in the existing indicator management platform:

[0074] The establishment of any business indicator needs to rely on the original or basic data of the business, so it is very important to comprehensively and completely concentrate the basic business data in advance, and process the business data from different system sources into commonly used factual data and dimensional data. Different from the construction of the traditional indicator data model, in addition to generating wide tables and summary tables required by some conventional business indicators, the present invention uses a large language model and knowledge base storage technology to generate different business indicator data semantic models based on factual data and dimensional data. In other words, when business personnel need to develop new business indicators, they first submit them to the large model, combine the indicator data semantic model stored in the knowledge base, and generate the relevant factual data and calculation methods that need to be processed for the indicator. On this basis, the wide table virtualization module submitted to the indicator calculation engine constructs a virtualized wide table, and finally the indicator calculation engine generates the calculation results. For virtualized wide tables that may be frequently reused, fixed cache technology is used to improve the indicator calculation performance.

[0075] Assume that the fact data and dimension data structure of a deposit business is as follows Figure 1As shown in the figure, in this example, branch_dim is the dimension information table of each branch of a bank. The table contains the ID, name, and address of the bank branch. money_dim is the dimension table of currency information supported by a bank, such as currency ID, currency name, country of currency, and currency symbol. customer_dim is the dimension table of information of all customers of a bank, such as customer ID, customer name, and age. product_dim is the dimension table of information of all products and services of a bank, such as bank product ID, product name, and product description. account_mgr_dim is the dimension information table of all account managers of the bank, such as account manager ID, account manager name, and account manager level. deposit_fact_2024 is the fact table of all deposits of the bank, which contains information about different deposit products deposited by all customers in different branches of the bank, as well as the account manager to whom the customer's service belongs.

[0076] In the existing deposit fact table, the business department can customize the corresponding business indicators, such as the amount of US dollar deposits of a certain customer in the most recent month. However, if the business department proposes to observe the amount of US dollar deposits of a certain institution's business manager that month, it is obviously impossible to directly calculate the indicator based on the conventional deposit fact table. Because in the deposit fact table depsoit_fact_2024 in this example, there is no definition of information related to the business manager, that is, account_mgr. Therefore, in the present invention, through a large language model, using the pre-stored business data semantic definition model and business prompt words, a virtual wide table structure and indicator calculation expression for the indicator "What is the amount of US dollar deposits of a certain institution's business manager that month" can be generated, such as Figure 2 shown.

[0077] (1) Figure 2 The specific meaning of the right part added in the code is as follows:

[0078] The enterprise's "complete business data" is processed into vectorized knowledge expression as knowledge through a large language model and stored as an indicator semantic model knowledge base. When a new business indicator is created, the prompt words in the prompt word library drive the large language model according to the new business indicator definition, and combine the indicator semantic model knowledge base to generate a virtualized wide table structure and a calculation expression for the corresponding indicator. Finally, the indicator calculation processing module generates the business indicator result based on the virtualized wide table structure and the indicator calculation expression.

[0079] (2) The calculation expression refers to the specific business indicator calculation expression, which is generated for the virtualized wide table. For example, the business department needs to obtain the operating indicator: What is the proportion of the corporate customer demand deposits of Changning Branch in the entire bank's deposits in August 2024? Therefore, based on the virtualized wide table virtual_deposit_fact data, the calculation expression of this indicator is:

[0080] --Get the total amount of current deposits of corporate customers of Changning Branch in August

[0081] SELECT SUM(deposit_amount)AS longning_corporate_current_deposit FROMvirtual_deposit_fact

[0082] WHERE branch = 'Changning Branch'

[0083] AND deposit_type = 'current deposit'

[0084] AND EXTRACT(YEAR FROM deposit_date)=2024

[0085] AND EXTRACT(MONTH FROM deposit_date)=8;

[0086] --Get the total deposit amount of the entire bank in August

[0087] SELECT SUM(deposit_amount)AS total_bank_deposit FROM virtual_deposit_factWHERE EXTRACT(YEAR FROM deposit_date)=2024AND EXTRACT(MONTH FROMdeposit_date)=8;

[0088] After getting the results of the above two queries, you can use the following expression to calculate the percentage: Percentage = (Total amount of current deposits of corporate customers of Changning Branch / Total amount of deposits of the entire bank in August) / (Total amount of deposits of the entire bank in August / Total amount of current deposits of corporate customers of Changning Branch) × 100%

[0089] The specific calculation steps are as follows:

[0090] 1. Execute the first query to obtain the total amount of current deposits of corporate customers of Changning Branch in August (recorded as A).

[0091] 2. Execute the second query to obtain the total deposit amount of the entire bank in August (recorded as B).

[0092] 3. Use the above formula to calculate the proportion = A / B X 100%

[0093] (3) virtual_deposit_fact is based on Figure 1 The virtual wide table is dynamically generated based on the definition of business indicators and other business data in the bank. The virtual wide table serves to calculate business indicators.

[0094] Figure 2 The "complete business data elements" of enterprises are processed into vectorized knowledge expressions through the big language model and stored as the indicator semantic model knowledge base. When a new business indicator is created, the big language model is driven by the prompt words in the prompt word library according to the new business indicator definition, and combined with the indicator semantic model knowledge base, a virtualized wide table structure and the calculation expression of the corresponding indicator are generated.

[0095] Finally, the indicator calculation processing module generates the business indicator result based on the virtualized wide table structure and indicator calculation expression. The process is as follows Figure 3 As shown:

[0096] 1) Platform business users define new business indicators based on market demand.

[0097] 2) The business indicator is submitted to the "big language model". The "big language model" analyzes and creates the data elements required to calculate the business indicator, and combines the definition of the indicator semantic model knowledge base, the prompt word library and other data elements of the enterprise to dynamically create a virtualized wide table structure and business indicator calculation expression.

[0098] 3) Based on the virtualized wide table structure and the business indicator calculation expression, the business indicator result is calculated through the business indicator calculation unit.

[0099] 4) Output the business indicator calculation results.

[0100] Therefore, by adopting the above-mentioned technical solution, the present invention intelligently and flexibly solves the problem of business application development and updating existing in the current indicator management platform.

[0101] The overall scheme of the present invention is as follows Figure 4 As shown:

[0102] 1) Collect relevant business data from various business systems of the enterprise.

[0103] 2) The collected business data is processed to form enterprise business fact data and related dimensional data.

[0104] 3) Use commonly used fact data and dimension data to generate wide tables and summary tables required for calculating the company's regular indicators.

[0105] 4) If the business indicators required by the business department are regular indicators and can be directly calculated through regular wide tables and summary tables, they are directly calculated through the indicator calculation processing unit and the indicator calculation results are output.

[0106] 5) If the business indicators required by the business department cannot be directly calculated through conventional wide tables and summary tables, new business indicators need to be created through the business department.

[0107] 6) Newly created business indicators are submitted to the "big language model". The "big language model" analyzes and creates the data elements required to calculate the business indicator, and combines the definition of the indicator semantic model knowledge base, the prompt word library and other data elements of the enterprise to dynamically create a virtualized wide table structure and the business indicator calculation expression.

[0108] 7) Based on the virtualized wide table structure and the business indicator calculation expression, the new business indicator result is calculated and output through the business indicator calculation unit.

[0109] 8) In order to accurately generate virtualized wide tables and indicator calculation expressions for new business indicators, the indicator semantic model knowledge base created in advance through the "big language model" is the basis.

[0110] In order to help the business departments of enterprises to create or optimize business observation indicators intelligently and flexibly, and improve the accuracy and timeliness of decision-making in the process of business development, the present invention implements the following calculation program working steps: Figure 5 Show:

[0111] In the calculation procedure work steps, the platform management user first starts the entire indicator semantic model knowledge base creation process, which is the basis of the entire program flow. At this stage, the platform management user needs to integrate various enterprise data elements to form an indicator semantic data model, and use the intelligence of the large language model to further assist the semantic understanding and generation of indicators. Ultimately, all this information and calculation logic are integrated into the indicator semantic model knowledge base in the form of vectorized data expression, providing a rich knowledge base for the entire platform, making the calculation of a wide range of business indicators more intelligent and flexible.

[0112] At the same time, platform business users start by defining new business indicators. Driven by business prompt words, they use large language models to understand the semantics of new business indicators, recall relevant indicator semantic knowledge, and generate virtualized wide table structures and indicator calculation expressions for business indicator calculation services. The indicator calculation unit is responsible for performing specific calculation tasks. Based on the virtualized wide table structure and indicator calculation expressions, it integrates enterprise data elements and generates the final business indicator calculation results.

[0113] The entire process demonstrates the creation of an indicator semantic data model to the generation and application of indicator calculation results. The two lines cooperate with each other to ensure the flexibility of business indicator calculation while improving the intelligence level of the entire indicator management platform.

[0114] Taking virtual_deposit_fact as an example, it is based on Figure 1 the tables and other business data in the bank. According to the definition of business indicators, a virtual wide table data structure is dynamically generated, as shown on the right side of Figure 2 . This virtual wide table serves for calculating business indicators.

[0115] The calculation expression refers to the specific business indicator calculation expression, which is generated corresponding to the virtualized wide table. For example, if the business part needs to obtain the indicator: What is the proportion of the current deposits of corporate customers in Changning Sub-branch in August 2024 in the total deposits of the entire bank in August? Therefore, based on the data of the virtualized wide table virtual_deposit_fact, the calculation expression for this indicator is:

[0116] -- Obtain the total current deposits of corporate customers in Changning Sub-branch in August

[0117] SELECT SUM(deposit_amount)AS longning_corporate_current_deposit FROMvirtual_deposit_factWHERE branch='Changning Sub-branch'

[0118] AND deposit_type='current'

[0119] AND EXTRACT(YEAR FROM deposit_date)=2024

[0120] AND EXTRACT(MONTH FROM deposit_date)=8;

[0121] -- Obtain the total deposits of the entire bank in August

[0122] SELECT SUM(deposit_amount)AS total_bank_deposit FROM virtual_deposit_factWHERE EXTRACT(YEAR FROM deposit_date)=2024

[0123] AND EXTRACT(MONTH FROM deposit_date)=8;

[0124] After getting the results of the above two queries, you can use the following expression to calculate the proportion:

[0125] Percentage = (Total amount of current deposits of corporate customers of Changning Branch / Total amount of deposits of the entire bank in August) / (Total amount of deposits of the entire bank in August / Total amount of current deposits of corporate customers of Changning Branch) × 100%

[0126] The specific calculation steps are as follows:

[0127] 1. Execute the first query to obtain the total amount of current deposits of corporate customers of Changning Branch in August (recorded as A).

[0128] 2. Execute the second query to obtain the total deposit amount of the entire bank in August (recorded as B).

[0129] 3. Use the above formula to calculate the proportion = A / B X 100%

[0130] The index calculation units are not the units of the present invention, they are common computer calculation units, such as CPU, GPU, etc. The present invention only submits the generated index calculation expression to them for calculation and obtains the result.

[0131] The indicator semantic data model is a structured model used to describe and define data indicators. It helps organizations clearly express the calculation logic, data sources, business meaning, etc. of business indicators.

[0132] The indicator semantic model usually contains the following key components: Indicator name: The name of the indicator, which is used to identify the indicator in reports and analysis. Business definition: The business logic description of the indicator, which explains the business meaning of the indicator and how it reflects business performance. Calculation logic: The calculation formula or algorithm of the indicator, including the mathematical expressions and logic used to calculate the indicator. Data source: The source of raw data required for indicator calculation, which may include database tables, data warehouses or other data stores. Dimensions and measures: Dimensions (such as time, location, product category, etc.) and measures (such as sales, profit, etc.) involved in indicator calculation. Filter conditions: Conditions used to limit the scope of indicator calculation, such as a specific time range, geographic location or business department. Update frequency: The update frequency of indicator data, such as daily, monthly or real-time updates.

[0133] Enterprise data elements refer to the key components that enterprises need when conducting business indicator management and analysis. These elements cover the entire process from the generation, storage, processing to analysis and application of indicator data. There are mainly the following parts: Enterprise data sources: Internal data sources: including data generated within the enterprise, such as sales data, financial reports, human resources information, inventory records, etc. External data sources: data from outside the enterprise, such as market research data, social media data, weather information, third-party data services, etc. Enterprise metadata: Business metadata: describes the business meaning and purpose of data, such as business terms, definitions, rules, etc. of data; Technical metadata: describes the technical details of data, such as data model, database structure, data storage location, data conversion rules, etc. Operational metadata: involves the management and maintenance process of data, such as the creation, update, access and maintenance of data records. Enterprise data types: Structured data: data stored in a relational database with a fixed format and pattern, such as customer information tables, transaction records, etc. Semi-structured data: data that does not fully conform to the relational database model, such as XML and JSON documents, which have a structure but are not as strict as structured data. Unstructured data: data without a fixed format, such as text files, pictures, videos, audio, etc.

[0134] Around the above Figure 5 The computer program working process steps of the invention are implemented as follows: Figure 6 As shown:

[0135] exist Figure 6 In the figure, the framework diagram of the indicator calculation intelligent processing engine of the present invention is presented, which includes the following modules: business indicator definition module, prompt word driving module, large language model module, indicator semantic data creation module, indicator semantic model generation module, indicator calculation unit module. Each module plays a different role in the whole framework. The following is a description of the functions of these modules:

[0136] 1. Business Indicator Definition Module: This module is the starting point of the entire framework. It is responsible for defining and creating new business indicators. Business indicators are key parameters for measuring the efficiency and effectiveness of enterprise operations. This module allows business departments to set or optimize business indicators based on the specific business development needs of the enterprise, ensuring the accuracy and applicability of the indicators.

[0137] 2. Prompt word driving module: This module is used to generate keywords or phrases driven by business indicators based on the definition of business indicators, which are used to guide the creation of subsequent indicator virtual wide tables and calculation expressions. This module uses the assembled prompt words, key information related to business indicators and indicator semantic model information to provide input for the large language model to generate indicator virtual wide table structures and indicator calculation expressions.

[0138] 3. Large language model module: This module uses advanced natural language processing technology to conduct in-depth analysis and understanding of input prompt words. The large language model can recognize and process complex language structures, thereby accurately capturing the semantic information of business indicators, indicator semantics, and enterprise-related data elements. It not only provides support for the generation of indicator semantic models, but also ensures that the platform can accurately reflect the calculation of business indicators.

[0139] 4. Module for creating indicator semantic data: With the help of the large language model module, this module is responsible for creating and maintaining indicator semantic data. These data are based on the enterprise data elements and enterprise indicator definitions, and include detailed descriptions of daily business indicators, calculation methods, and related parameters. With these data, the platform can more accurately calculate indicators within the enterprise.

[0140] 5. Indicator semantic model generation module: This module generates a specific indicator semantic model based on the created indicator semantic data, and expresses the model data in a vectorized form. The model not only includes the definition of enterprise-related data elements, but also includes the definition of business indicators, and the logic and rules required to calculate these indicators. This module ensures the accuracy and consistency of enterprise indicator calculations.

[0141] 6. Index calculation unit module: This module is responsible for performing the actual index calculation. It uses the index semantic model to generate a virtual wide table structure and index calculation expression, combined with actual business data, to perform calculations and generate results.

[0142] The design of the entire framework reflects the complete process from business indicator definition to final calculation results. Each module is closely connected and interdependent, jointly ensuring the accuracy and flexibility of indicator calculation. Through this design, enterprises can better understand, utilize and create new business indicators based on the ever-changing market, thereby optimizing the decision-making process and improving operational efficiency.

[0143] In the present invention, platform management users integrate various enterprise data elements by "creating indicator semantic data modules" and use the intelligence of "large language model modules" to further assist in the semantic understanding and generation of enterprise data and indicators. Finally, through the "indicator semantic model generation module", all this information and calculation logic are stored in the indicator semantic model knowledge base in the form of structured and vectorized data expressions, pre-building a rich enterprise indicator knowledge base for the entire platform, making the calculation of a wider range of business indicators more intelligent and flexible.

[0144] At the same time, in the present invention, the platform business users start by defining new business indicators using the "business indicator definition module". Driven by the "business prompt word module", the semantics of the new business indicators are understood through the "big language model", and the semantic knowledge of related enterprise indicators is recalled. On this basis, the "virtualized wide table structure" and "indicator calculation expression" for new business indicator calculation services are generated. The "indicator calculation unit module" is responsible for performing specific calculation tasks. On the basis of the virtualized wide table structure and indicator calculation expression, the corresponding enterprise data elements are integrated to generate the final business indicator calculation results.

[0145] The hardware structure of the computer device of the present invention is as follows Figure 7 As shown:

[0146] Among them, the business indicator definition module, the enterprise language model module, the indicator semantic model generation module and the indicator calculation unit module constitute the core foundation of the present invention. In the present invention, Figure 7 Each module in the computer hardware system corresponds to a different server cluster. The enterprise big model language module runs on the GPU server cluster, the management database cluster runs on the MySQL server cluster, and the enterprise data warehouse cluster runs on the Huawei HiSilicon big data platform. Other corresponding modules run on the K8S server cluster.

[0147] The business indicator definition module, prompt word driving module, indicator semantic data creation module and indicator calculation unit module complete the definition of business indicators, the definition of indicator semantic data and the implementation of indicator calculation through Java and corresponding Springboot development components. In other modules of the present invention, the communication between modules can be completed by calling the interface API provided by the above modules.

[0148] The enterprise language model module and the indicator semantic generation module are implemented based on Python and related components. Through these two modules, the server cluster completes the semantic information of the business indicators involved in the present invention, the indicator semantics and the enterprise-related data element information are accurately captured and understood, and a specific indicator semantic model is generated to express the model data in a vectorized form. This ensures the accuracy and consistency of the enterprise indicator calculation.

[0149] The management database cluster is the storage basis for the indicator semantic data involved in the present invention. The structured data involved in the platform of the present invention is stored in a database cluster manner, and a MySQL cluster server is used as a support. However, the database clusters in the present invention are not limited to MySQL. The file server cluster is mainly used in the present invention to store virtualized wide table structures and indicator calculation expressions, and then provide file services for the indicator calculation unit module. The enterprise data warehouse cluster exists in an optional manner in the present invention to store various types of enterprise data elements. If the data warehouse cluster already exists in the enterprise, it can be directly connected to the present invention in the form of data consumption.

[0150] Compared with the existing technology implementation, the present invention has the following advantages:

[0151] First, the present invention effectively responds to changes in business indicators caused by rapid changes in the business environment by introducing a dynamic and flexible indicator data model. In traditional indicator management platforms, the fixed and rigid data model leads to limited responsiveness to changes in corporate market dynamics and business indicator requirements. The present invention allows instant adjustment and dynamic updating of indicator data models, allowing enterprises to quickly adapt to a wider range of business logic and ensure the timeliness and accuracy of business indicator analysis results. In the present invention, the indicator data model no longer uses an unchanging or relatively fixed data model like the prior art implementation method. Figure 8 As shown in the figure, the database table structure and calculation expressions that all business indicator calculations in the prior art rely on are relatively fixed.

[0152] In the present invention, the indicator data model will dynamically construct a virtualized wide table and indicator calculation expression based on the enterprise indicator semantic model and the new enterprise indicator definition, based on the natural language intent recognition and generation capabilities of the enterprise large language model, and submit it to the indicator calculation engine to generate results. Figure 6 As shown, based on the existing platform, the present invention adds a business indicator definition module, a prompt word driving module, a large language model module, a module for creating indicator semantic data, an indicator semantic model generation module, and an indicator calculation unit module. Thus, according to the business indicator requirements, the "virtualized wide table structure" and "indicator calculation expression" are dynamically generated. The definition and calculation of business indicators are no longer tightly coupled and dependent on pre-defined database tables.

[0153] Secondly, the present invention simplifies and reconstructs the indicator processing flow, reduces the dependence on professionals, and improves the intelligence and automation level of flexible definition and calculation of business indicators. Under the existing platform, in the face of major changes in business scenarios, enterprises usually need to rely on ETL engineers to reprocess data or define data models. This process is not only time-consuming and labor-intensive, but may also miss key business opportunities. The present invention uses a dynamic and intelligent way to implement the indicator data model. On the basis of various data elements of the enterprise, combined with the pre-generated enterprise indicator semantic model library of the present invention, the natural language intent recognition and generation capabilities of the enterprise's large language model are used to automatically realize the calculation of business indicators. In this process, many traditional tasks that rely on professionals are replaced by the intelligence of the enterprise's large language model. Therefore, while accelerating the response speed of enterprises to market changes, it also greatly improves the generalization processing capabilities of responding to enterprise business indicators.

[0154] In addition, the present invention enhances the ability to understand and express the business logic of an enterprise by pre-building an enterprise indicator semantic data model. This model can not only describe business needs more accurately, but also provide reusable business indicator analysis perspectives and elements. The indicator semantic data model of the present invention is constructed on the basis of various data elements of the enterprise. Therefore, compared with traditional wide tables and summary tables, the enterprise indicator semantic data model can adapt to complex business scenarios more flexibly and help enterprises better understand market dynamics and business trends.

[0155] Finally, the present invention also improves the intelligence level of the platform by introducing large language models and natural language processing technology. This allows non-professionals to easily define and update indicators, lowers the threshold for using the platform, and also improves the satisfaction of business departments' real-time data needs. This intelligent data processing method not only reduces the workload of professionals, but also reduces the difficulty of using the entire platform.

[0156] In summary, the present invention flexibly solves the pain points of the current system platform in terms of indicator processing, data response, operation complexity and computing efficiency through intelligent, automated and semantic technical means. It provides enterprises with a more flexible, efficient and intelligent indicator management application solution, helping enterprises to maintain competitiveness in a rapidly changing business environment and seize key business opportunities.

[0157] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. An indicator calculation and processing engine based on an indicator wide table virtualization data model, characterized in that: include: Business indicator definition module: responsible for defining and creating new business indicators. This module allows business departments to set or optimize business indicators based on the specific business development needs of the enterprise, ensuring the accuracy and applicability of the indicators; Prompt word driving module: It is used to generate keywords or phrases driven by business indicators based on the definition of business indicators, and to guide the creation of subsequent virtual wide tables and calculation expressions of indicators. Through the assembled prompt words, key information related to business indicators and indicator semantic model information are used to provide input for the large language model to generate virtual wide table structures and indicator calculation expressions of indicators. Large language model module: uses natural language processing technology to conduct in-depth analysis and understanding of input prompt words; it can identify and process complex language structures, thereby accurately capturing the semantic information of business indicators, indicator semantics, and enterprise-related data elements, which not only provides support for the generation of indicator semantic models, but also ensures that the platform can accurately reflect the calculation of business indicators; Indicator semantic data creation module: Based on the large language model module, it is responsible for creating and maintaining indicator semantic data; the indicator semantic data is the basis for enterprise data elements and enterprise indicator definitions, including detailed descriptions of daily business indicators, calculation methods and related parameters; with these data, the platform can more accurately calculate indicators within the enterprise; Indicator semantic model generation module: Generates a specific indicator semantic model based on the created indicator semantic data, and expresses the model data in a vectorized form; the indicator semantic model not only includes the definition of enterprise-related data elements, but also includes the definition of business indicators and the logic and rules required to calculate these indicators; Indicator calculation unit module: responsible for performing actual indicator calculations, generating virtual wide table structures and indicator calculation expressions through indicator semantic models, and performing calculations and generating results based on actual business data; Platform management users create indicator semantic data modules, integrate various enterprise data elements, and use the intelligence of the large language model module to further assist in the semantic understanding and generation of enterprise data and indicators. Finally, the indicator semantic model generation module stores all this information and calculation logic in the indicator semantic model knowledge base in the form of structured and vectorized data expressions, pre-building a rich enterprise indicator knowledge base for the entire platform, making the calculation of a wider range of business indicators more intelligent and flexible. The platform business users start by defining new business indicators using the business indicator definition module. Driven by the business prompt word module, the semantics of the new business indicators are understood through the large language model, and the relevant enterprise indicator semantic knowledge is recalled. On this basis, a new virtualized wide table structure and indicator calculation expression for the business indicator calculation service are generated. The indicator calculation unit module performs specific calculation tasks, and on the basis of the virtualized wide table structure and indicator calculation expression, the corresponding enterprise data elements are integrated to generate the final business indicator calculation results.

2. The indicator calculation and processing engine based on the indicator wide table virtualization data model according to claim 1 is characterized in that: The computer hardware system of the index calculation and processing engine includes a GPU server cluster and a K8S server cluster; The large language model module runs on a GPU server cluster; The business indicator definition module, prompt word driving module, indicator semantic model generation module, indicator semantic data creation module and indicator calculation unit module are respectively run on the K8S server cluster, and the definition of business indicators, the driving of prompt words, the generation of indicator semantic models, the definition of indicator semantic data and the implementation of indicator calculation are completed through Java and corresponding Springboot development components, and the communication between modules is completed by calling the API interface provided by each module; The large language model module and the indicator semantic generation module complete the accurate capture and understanding of the semantic information of the business indicators involved, the indicator semantics and the enterprise-related data element information, and generate a specific indicator semantic model, expressing the model data in a vectorized form, ensuring the accuracy and consistency of the enterprise indicator calculation.

3. The indicator calculation and processing engine based on the indicator wide table virtualization data model according to claim 2 is characterized in that: A file server cluster also runs on the K8S server cluster; the file server cluster is used to store virtualized wide table structures and indicator calculation expressions, and then provide file services for the indicator calculation unit module.

4. The indicator calculation and processing engine based on the indicator wide table virtualization data model according to claim 2 is characterized in that: An enterprise data warehouse cluster is also running on the K8S server cluster, and the enterprise data warehouse cluster is used to store various enterprise data elements; if the enterprise data warehouse cluster already exists in the enterprise, it is directly connected to the indicator calculation and processing engine in the form of data consumption.

5. The indicator calculation and processing engine based on the indicator wide table virtualization data model according to claim 2 is characterized in that: A management database cluster is also running on the K8S server cluster; the management database cluster uses a database cluster to store structured data related to the platform of the present invention.

6. The indicator calculation and processing engine based on the indicator wide table virtualization data model according to claim 5 is characterized in that: The management database cluster uses MySQL cluster server as support.

7. The implementation method based on the indicator wide table virtualized data model is implemented by using the indicator calculation processing engine based on the indicator wide table virtualized data model as described in any one of claims 1 to 7, characterized in that: The steps include: S1. Create an indicator semantic model knowledge base through a large language model and collect business data from various business systems of the enterprise; S2. The collected business data is processed to form enterprise business fact data and related dimension data; S3. Use commonly used fact data and dimension data to generate wide tables and summary tables required for calculating the company's regular indicators; S4. If the business indicator required by the business department is a regular indicator and can be directly calculated through regular wide tables and summary tables, it is directly calculated through the indicator calculation processing unit and the indicator calculation result is output; if the business indicator required by the business department cannot be directly calculated through regular wide tables and summary tables, a new business indicator needs to be created through the business department; S5. Submit the newly created business indicator to the big language model. Since the big language model analyzes and creates the data elements required to calculate the business indicator, and combines the definition of the indicator semantic model knowledge base, prompts the vocabulary and other data elements of the enterprise, it dynamically creates a virtualized wide table structure and the business indicator calculation expression. S6. Based on the virtualized wide table structure and the business indicator calculation expression, a new business indicator result is calculated and output through a business indicator calculation unit.

8. The method for implementing the virtualized data model based on the indicator wide table according to claim 7 is characterized in that: The implementation method is also divided into platform management user steps and platform business user steps: Steps for platform business users: M1. The platform management user starts the entire indicator semantic model knowledge base creation process: the platform management user integrates various enterprise data elements, creates and forms the indicator semantic data model; M2, using the intelligence of large language models to further assist in the semantic understanding and generation of indicators; M3. All this information and calculation logic are integrated into the indicator semantic model knowledge base in the form of vectorized data expression, providing a rich knowledge base for the entire platform, making the calculation of a wide range of business indicators more intelligent and flexible; Steps for platform business users: P1. Platform business users start by defining new business indicators. Driven by the prompt word driving module, they use the large language model to understand the semantics of new business indicators, recall relevant indicator semantic knowledge, and generate the virtualized wide table structure and indicator calculation expression of the business indicator calculation service; P2, the indicator calculation unit module is responsible for executing specific calculation tasks. Based on the virtualized wide table structure and indicator calculation expression, it integrates enterprise data elements and generates the final enterprise indicator settlement results; The virtualized wide table structure refers to a virtual wide table data structure dynamically generated according to the definition of business indicators; the indicator calculation expression is a specific business indicator calculation expression, which is generated corresponding to the virtualized wide table.

9. The method for implementing the virtualized data model based on the indicator wide table according to claim 8 is characterized in that: The indicator semantic data model is used to describe and define a structured model of data indicators, helping organizations to clearly express the calculation logic, data sources, and business meanings of business indicators; the indicator semantic data model includes: Business definition: a description of the business logic of the indicator, explaining the business meaning of the indicator and how it reflects business performance; Calculation logic: the calculation formula or algorithm of the indicator, including the mathematical expression and logic used to calculate the indicator; Data source: the source of raw data required for indicator calculation, including database tables, data warehouses or other data storage; Dimensions and measures: Dimensions and measures involved in the calculation of indicators. Dimensions include time, location, and product category; measures include sales and profits; Filters: Conditions used to limit the scope of metric calculations, including specific time ranges, geographic locations, or business units; Update frequency: The update frequency of indicator data, including daily, monthly or real-time updates.

10. The method for implementing the virtualized data model based on the indicator wide table according to claim 8 is characterized in that: The enterprise data elements refer to all business-related data of the enterprise, which are the key components required by the enterprise when conducting business indicator management and analysis, covering the entire process from the generation, storage, processing to analysis and application of indicator data: The enterprise data elements include: Enterprise metadata: includes business metadata, technical metadata and operational metadata; business metadata is used to describe the business meaning and purpose of data, including business terms, definitions and rules of data; technical metadata is used to describe the technical details of data, including data model, database structure, data storage location and data conversion rules; operational metadata involves the management and maintenance process of data, including data creation, update, access and maintenance records; Enterprise data types: including structured data, semi-structured data and unstructured data; the structured data is data stored in a relational database with a fixed format and pattern, including customer information tables and transaction records; the semi-structured data is data that does not fully conform to the relational database model, including XML and JSON documents, which have a structure but are not as strict as structured data; the unstructured data does not have a fixed format, including text files, pictures, videos and audio.

Images