Data warehouse structure generation method and device, computer equipment and storage medium

Abstract

The invention relates to a data warehouse structure generation method and device, computer equipment, a storage medium and a computer program product. The method can be applied to artificial intelligence, such as a scene of constructing a data warehouse structure through an intelligent question and answer task. The method comprises the following steps: generating first inquiry information according to knowledge data of a service, and inquiring entity data based on the first inquiry information; generating second inquiry information based on the entity data and the service demand information, and replying the second inquiry information to obtain key entity data corresponding to the service demand information; third inquiry information is generated based on the key entity data, the business demand information and the entity constraint information, a question and answer task is executed based on the third inquiry information, and table structures of different levels and the hierarchical relation between the table structures are obtained; and constructing a data warehouse structure according to the table structures of different hierarchies and the hierarchical relationship between the table structures. By adopting the method, the efficiency of constructing the data warehouse structure can be improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, apparatus, computer equipment, storage medium, and computer program product for generating a data warehouse structure. Background Art

[0002] A data warehouse (DW) is a subject-oriented, integrated, non-volatile, and time-variant data collection used to support management decisions and global information sharing. The data warehouse structure has a significant impact on its performance.

[0003] In related technologies, data warehouse structures are usually designed by experts based on business requirements, resulting in low efficiency in constructing data warehouse structures. Summary of the Invention

[0004] Based on this, it is necessary to provide a method, device, computer equipment, computer-readable storage medium and computer program product for generating a data warehouse structure to address the above technical problems, which can improve the efficiency of constructing a data warehouse structure.

[0005] In a first aspect, the present application provides a method for generating a data warehouse structure. The method comprises:

[0006] Generate a first query message based on the business knowledge data, and query the business entity data based on the first query message; generate a second query message based on the business entity data and business demand information, and reply to the second query message to obtain the key entity data corresponding to the business demand information; generate a third query message based on the key entity data, business demand information and entity constraint information, and perform question-and-answer tasks based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures; construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between table structures.

[0007] In a second aspect, the present application also provides a device for generating a data warehouse structure. The device includes:

[0008] An entity data determination module, configured to generate first query information based on the knowledge data of the business, and query the entity data of the business based on the first query information;

[0009] A key entity data determination module is used to generate second query information based on the business entity data and business requirement information, and to respond to the second query information to obtain key entity data corresponding to the business requirement information;

[0010] A table structure and hierarchical relationship determination module is used to generate third query information based on key entity data, business requirement information, and entity constraint information, and perform a question-answering task based on the third query information to obtain table structures at different levels and the hierarchical relationships between the table structures;

[0011] The data warehouse structure construction module is used to construct a data warehouse structure corresponding to business demand information based on table structures at different levels and the hierarchical relationships between table structures.

[0012] In some embodiments, in some embodiments, the key entity data determination module is further used to functionally split the business requirement information included in the second query information to obtain each sub-requirement information; and select the key entity data corresponding to each sub-requirement information from the entity data included in the second query information.

[0013] In some embodiments, the table structure and hierarchical relationship determination module is also used to select key entity data of different levels from the key entity data included in the third query information based on business demand information, and determine the dependency relationships between the key entity data of different levels; based on the key entity data of different levels, the dependency relationships and the entity constraint information included in the third query information, determine the table structures of different levels and the hierarchical relationships between the table structures.

[0014] In some embodiments, the table structure and hierarchical relationship determination module is further used to determine logical requirement information and display requirement information based on business requirement information; among the key entity data included in the third query information, the first key entity data belonging to the source level and the second key entity data belonging to the aggregation level are selected according to the logical requirement information, and a first dependency relationship between the first key entity data and the second key entity data is determined; among the key entity data, the third key entity data belonging to the application level is selected according to the display requirement information, and a second dependency relationship between the first key entity data, the second key entity data, and the third key entity data is determined.

[0015] In some embodiments, the table structure includes a first table structure, a second table structure, and a third table structure; the hierarchical relationship includes a first hierarchical relationship and a second hierarchical relationship; the table structure and hierarchical relationship determination module is further used to construct a first table structure belonging to the source level and a second table structure belonging to the aggregation level based on the first key entity data, the second key entity data, the first dependency relationship, and the entity constraint information included in the third query information; determine the first hierarchical relationship between the first table structure and the second table structure based on the first dependency relationship; construct a third table structure belonging to the application level based on the first key entity data, the second key entity data, the third key entity data, the second dependency relationship, and the entity constraint information; determine the second hierarchical relationship between the first table structure, the second table structure, and the third table structure based on the second dependency relationship.

[0016] In some embodiments, the table structure includes a first table structure, a second table structure, and a third table structure; the table structure and hierarchical relationship determination module is further used to determine logical requirement information and display requirement information based on business requirement information; among the key entity data included in the third query information, the first key entity data of the source level is selected according to the logical requirement information; based on the first key entity data and the entity constraint information included in the third query information, a first table structure belonging to the source level is constructed; based on the first table structure, the logical requirement information, and the display requirement information, a second table structure belonging to the aggregation level and a third table structure belonging to the application level are constructed, and the hierarchical relationship between the first table structure, the second table structure, and the third table structure is determined.

[0017] In some embodiments, the hierarchical relationship includes a first hierarchical relationship and a second hierarchical relationship; the table structure and hierarchical relationship determination module is further used to select a first candidate table structure in the first table structure based on logical requirement information; perform aggregation processing on the first candidate table structure to obtain a second table structure belonging to the aggregation level; select a second candidate table structure from the first table structure and the second table structure based on display requirement information; perform data extraction processing on the second candidate table structure to obtain a third table structure belonging to the application level; determine the first hierarchical relationship between the first candidate table structure and the second table structure, and determine the second hierarchical relationship between the second candidate table structure and the third table structure.

[0018] In some embodiments, the entity data determination module is also used to respond to the first query information through the big model to obtain the entity data of the business; the key entity data determination module is also used to respond to the second query information through the big model to obtain the key entity data corresponding to the business demand information; the table structure and hierarchical relationship determination module is also used to execute the question-and-answer task corresponding to the third query information through the big model to obtain the table structures at different levels and the hierarchical relationships between the table structures.

[0019] In some embodiments, the data warehouse structure generation device also includes: a first large model training module, which is used to generate first sample query information based on sample knowledge data, and process the first sample query information through the initial large model to obtain training entity data; generate second sample query information based on the training entity data and sample requirement information, and process the second sample query information through the initial large model to obtain training key entity data; generate third sample query information based on the training key entity data, sample requirement information and sample entity constraint information, and process the third sample query information through the initial large model to obtain training table structures at different levels and hierarchical relationships between training table structures; construct a training data warehouse structure based on training table structures at different levels and hierarchical relationships between training table structures; adjust the parameters of the initial large model based on the training data warehouse structure and the label data warehouse structure until the initial large model converges to obtain a large model.

[0020] In some embodiments, the data warehouse structure generation device also includes: a data warehouse structure verification module, which is used to generate a fourth query information for data warehouse verification based on the data warehouse structure to be processed; and the fourth query information is responded to and processed through the large model to obtain the abnormal table structure in the data warehouse structure to be processed.

[0021] In some embodiments, the data warehouse structure generation device also includes: a second large model training module, which is used to modify the label data warehouse structure to obtain the abnormal data warehouse structure; generate a fourth sample query information based on the abnormal data warehouse structure; reply to the fourth sample query information through the large model to obtain a training abnormality table structure; determine the label abnormality table structure based on the difference between the label data warehouse structure and the abnormal data warehouse structure; adjust the parameters of the large model according to the training abnormality table structure and the label abnormality table structure until the parameters of the large model converge to obtain an updated large model.

[0022] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are performed:

[0023] Generate a first query message based on the business knowledge data, and query the business entity data based on the first query message; generate a second query message based on the business entity data and business demand information, and reply to the second query message to obtain the key entity data corresponding to the business demand information; generate a third query message based on the key entity data, business demand information and entity constraint information, and perform question-and-answer tasks based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures; construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between table structures.

[0024] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps:

[0025] Generate a first query message based on the business knowledge data, and query the business entity data based on the first query message; generate a second query message based on the business entity data and business demand information, and reply to the second query message to obtain the key entity data corresponding to the business demand information; generate a third query message based on the key entity data, business demand information and entity constraint information, and perform question-and-answer tasks based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures; construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between table structures.

[0026] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the following steps:

[0027] Generate a first query message based on the business knowledge data, and query the business entity data based on the first query message; generate a second query message based on the business entity data and business demand information, and reply to the second query message to obtain the key entity data corresponding to the business demand information; generate a third query message based on the key entity data, business demand information and entity constraint information, and perform question-and-answer tasks based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures; construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between table structures.

[0028] The above-mentioned data warehouse structure generation method, device, computer equipment, storage medium and computer program product, through multiple intelligent question-answering tasks, gradually determine the entity data of the business, extract the key entity data of the function corresponding to the business requirement information, determine the different levels of table structures and the hierarchical relationship between the table structures that meet the business requirement information and entity constraint information, and then can construct the data warehouse structure according to the different levels of table structures and the hierarchical relationship between the table structures. Compared with the manual construction of the data warehouse structure corresponding to the business requirement information by experts in the business field, it can realize the automatic construction of the data warehouse structure corresponding to the business requirement information, thereby improving the efficiency of constructing the data warehouse structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is an application environment diagram of a method for generating a data warehouse structure in one embodiment;

[0030] Figure 21 is a flow chart of a method for generating a data warehouse structure in one embodiment;

[0031] Figure 3 A schematic diagram of table structures at various levels and hierarchical relationships between table structures in one embodiment;

[0032] Figure 4 A schematic diagram of a method for generating a data warehouse structure using a large model in one embodiment;

[0033] Figure 5 A schematic diagram of a process for adjusting parameters of an initial large model to obtain a large model in one embodiment;

[0034] Figure 6 A schematic diagram of determining a loss value used to adjust parameters of an initial large model in one embodiment;

[0035] Figure 7 A schematic diagram of determining a loss value used to adjust parameters of an initial large model in another embodiment;

[0036] Figure 8 A schematic diagram of a process for verifying a data warehouse structure to be processed in one embodiment;

[0037] Figure 9 A schematic diagram of a process for adjusting parameters of a large model to obtain an updated large model in one embodiment;

[0038] Figure 10 A schematic diagram of key entities at each level and dependency relationships between key entities in one embodiment;

[0039] Figure 11 A schematic diagram of a data warehouse structure for building a transaction business in one embodiment;

[0040] Figure 12 A schematic diagram of verifying and improving the structure of a warehouse of data to be processed for financial services in one embodiment;

[0041] Figure 13 is a flowchart of a method for generating a data warehouse structure in another embodiment;

[0042] Figure 14 is a structural block diagram of a device for generating a data warehouse structure in one embodiment;

[0043] Figure 15 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0045] Artificial Intelligence (AI) refers to the theories, methods, techniques, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, to perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that seeks to understand the essence of intelligence and produce new intelligent machines that can respond in a manner similar to human intelligence. AI also involves studying the design principles and implementation methods of various intelligent machines, enabling them to possess the capabilities of perception, reasoning, and decision-making.

[0046] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, pre-trained models, operating / interaction systems, and mechatronics. Pre-trained models, also known as large models or basic models, can be fine-tuned and widely applied to downstream tasks across various AI disciplines. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0047] With the research and advancement of artificial intelligence technology, artificial intelligence technology has been studied and applied in many fields, such as common smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, unmanned driving, autonomous driving, drones, digital twins, virtual humans, robots, artificial intelligence generated content (AIGC), conversational interaction, smart medical care, smart customer service, game AI, etc. It is believed that with the development of technology, artificial intelligence technology will be applied in more fields and play an increasingly important role.

[0048] The data warehouse structure generation method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store data that server 104 needs to process. The data storage system can be integrated with server 104, or placed on a cloud or other network server. The method for generating a data warehouse structure can be executed by terminal 102 or server 104, or by terminal 102 and server 104 working together.

[0049] Taking the method of generating a data warehouse structure as an example, the terminal 102 generates a first query information based on the knowledge data of the business, and queries the entity data of the business based on the first query information; the terminal 102 generates a second query information based on the entity data and business demand information of the business, and responds to the second query information to obtain key entity data corresponding to the business demand information; the terminal 102 generates a third query information based on the key entity data, business demand information and entity constraint information, and performs a question-and-answer task based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures; the terminal 102 constructs a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between table structures.

[0050] The terminal 102 may be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smart watch, IoT device, or portable wearable device. IoT device may be a smart speaker, smart TV, smart air conditioner, or smart car device. Portable wearable device may be a smart watch, smart bracelet, or head-mounted device.

[0051] The server 104 may be an independent physical server or a service node in a blockchain system, where each service node in the blockchain system forms a peer-to-peer network.

[0052] In addition, server 104 can also be a server cluster composed of multiple physical servers, and can be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0053] The terminal 102 and the server 104 can be connected via Bluetooth, USB (Universal Serial Bus) or network communication connection methods, and this application does not limit this.

[0054] In some embodiments, as Figure 2 As shown, a method for generating a data warehouse structure is provided. Figure 1 The server or terminal in the Figure 1 The server and terminal in the embodiment cooperate to execute the method, and the method is described as follows: the method is executed by a computer device, and the computer device can be Figure 1 The server or terminal in the embodiment includes the following steps:

[0055] Step 202: Generate first query information based on the knowledge data of the business, and query entity data of the business based on the first query information.

[0056] Among them, business knowledge data refers to business-related knowledge content, which may include business definitions, business usage scenarios, and key technical points of the business. In actual applications, the business may be payment business, search business, resource transfer business, etc.

[0057] The first query information is used to query the entity data included in the knowledge data. For example, the first query information is used to indicate what entity data can be extracted based on the knowledge data.

[0058] Business entity data includes entities related to the business, entity attributes of entities, and entity relationships; entity attributes can represent the characteristics or properties of the entity. For example, if the entity is an organization, the entity attributes of the organization may include: number of employees, registered capital, etc.

[0059] Entity relationships are used to represent the relationship between entities. For example, entity E1 is an account entity, entity E2 is a user entity, and the entity relationship of entity E1 includes the ownership relationship between entity E1 and entity E2.

[0060] Specifically, the computer device acquires knowledge data of the business, embeds the acquired knowledge data into a query template for querying entity data, and obtains first query information; based on the first query information, the entity data of the business is queried, and the first query information can be processed by a large model to obtain the entity data of the business; the large model is used to process intelligent question-answering tasks. In practical applications, the large model can be implemented by a GPT model, which is a generative pre-trained model (Generative Pre-Trained Transformer).

[0061] Exemplarily, a question (first inquiry information) for inquiring about entity data of the business is generated based on the knowledge data of the business, and the question is input into the big model. The answer output by the big model is the entity data of the business.

[0062] Step 204 : Generate second query information based on the business entity data and business requirement information, and perform reply processing on the second query information to obtain key entity data corresponding to the business requirement information.

[0063] Among them, business demand information is used to represent the functions that the business needs to implement, which can be functions used by users. For example, business demand information includes querying annual income, which means that the function of querying annual income needs to be implemented; business demand information can be set according to the actual needs of the business.

[0064] The second query information is used to query the key entity data in the entity data for realizing the function corresponding to the business requirement information; the key entity data may be at least part of the entity data in the entity data, specifically the key entities corresponding to the business requirement information, the entity attributes and entity relationships of the key entities; the key entities corresponding to the business requirement information may be at least part of the entities corresponding to the business.

[0065] Specifically, the computer device embeds the business entity data and business requirement information into a query template for querying key entity data to obtain second query information.

[0066] In one implementation, reply processing is performed based on the second query information to obtain key entity data for implementing the function corresponding to the business requirement information. The second query information can be processed through a large model to obtain key entity data corresponding to the business requirement information.

[0067] Exemplarily, the entity data of the payment business includes: user entity data, account entity data, income entity data, expenditure entity data, order entity data and biometric information entity data; the business requirement information is to query annual income; second query information is generated based on the entity data and business requirement information of the payment business, and the second query information is processed by the big model, and the key entity data obtained are: user entity data, account entity data and income entity data.

[0068] In another implementation, reply processing is performed based on the second query information to obtain key entity data for implementing the function corresponding to the business requirement information. This can be to obtain the second query information including the business requirement information, determine the key entities involved in the business requirement information, and select the key entity data of the key entity from the entity data.

[0069] Step 206 : Generate third query information based on the key entity data, business requirement information, and entity constraint information, and perform a question-answering task based on the third query information to obtain table structures at different levels and hierarchical relationships between the table structures.

[0070] Among them, the entity constraint information is used to reflect the restriction conditions of the entity, specifically the restriction conditions of the entity corresponding table structure; for example, the number of account entities associated with the user entity does not exceed the preset number; the attributes of the account entity include the preset area.

[0071] The third query information is used to inquire about the table structures of key entities at each level involved in implementing the functions corresponding to business requirements and the hierarchical relationships between the table structures; the table structure of the key entity can be the data structure of the SQL table corresponding to the key entity, the fields in the table structure correspond to the entity attributes of the key entity, and the relationship keys in the table structure represent the entity relationships between the key entities; SQL (Structured Query Language) is a structured query language, and the SQL table describes the table structure through SQL statements.

[0072] Table structures at different levels include, but are not limited to, source-level table structures, aggregation-level table structures, and application-level table structures, and may also include storage-level table structures.

[0073] The hierarchical relationship between table structures refers to the association relationship between table structures at different levels. It can include the association relationship between the table structure at the source level and the table structure at the aggregation level, the association relationship between the table structure at the source level, the table structure at the aggregation level, and the table structure at the application level, the association relationship between the table structure at the source level and the table structure at the storage level, etc. It should be noted that the existence of an association relationship between table structures at different levels means that the table structure at one level is determined based on the table structure at another level.

[0074] Specifically, the computer device embeds the key entity data, business requirement information, and entity constraint information into the query table structure and the query template of the hierarchical relationship to obtain the third query information.

[0075] In one implementation, a question-answering task is performed based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures. This can be achieved by processing the third query information through a large model to obtain table structures at different levels and hierarchical relationships between table structures.

[0076] In another implementation, a question-and-answer task is performed based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures. This can be done by determining key entity data at different levels and entity relationships between key entity data at different levels in the key entity data included in the third query information based on business requirement information. Based on the key entity data and entity relationships at different levels, table structures at different levels are constructed and the hierarchical relationships between the table structures are determined.

[0077] In another implementation, a question-and-answer task is performed based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures. This may be to determine source-level entity data in key entity data based on business requirement information, construct a source-level table structure based on the source-level entity data, construct an aggregation-level table structure and an application-level table structure based on the source-level table structure and business requirement information, and determine the hierarchical relationship between the source-level table structure, the aggregation-level table structure, and the application-level table structure.

[0078] Step 208: construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between the table structures.

[0079] Among them, the data warehouse structure can represent table structures at different levels and the hierarchical relationships between table structures at different levels.

[0080] Specifically, the computer equipment determines the sub-warehouse structures at each level included in the data warehouse structure based on the table structures at each level, determines the hierarchical relationship between the sub-warehouse structures at each level based on the hierarchical relationship between the table structures at each level, and constructs the data warehouse structure based on the sub-warehouse structures at each level and the hierarchical relationship between the sub-warehouse structures at each level.

[0081] Exemplarily, the table structures at different levels include the table structure at the source level, the table structure at the aggregation level, and the table structure at the application level. The sub-warehouse structure at the source level is determined based on the table structure at the source level, the sub-warehouse structure at the aggregation level is determined based on the table structure at the aggregation level, and the sub-warehouse structure at the application level is determined based on the table structure at the application level. Based on the hierarchical relationship between the table structure at the source level and the table structure at the aggregation level, the hierarchical relationship between the sub-warehouse structure at the source level and the sub-warehouse structure at the aggregation level is established. Based on the hierarchical relationship between the table structure at the source level and the table structure at the application level, the hierarchical relationship between the sub-warehouse structure at the source level and the sub-warehouse structure at the application level is established. Moreover, based on the hierarchical relationship between the table structure at the aggregation level and the table structure at the application level, the hierarchical relationship between the sub-warehouse structure at the aggregation level and the sub-warehouse structure at the application level is established.

[0082] In actual applications, the hierarchical relationship between the sub-warehouse structure at the source level and the sub-warehouse structure at the application level can be represented by relationship lines. Similarly, the hierarchical relationship between the sub-warehouse structure at the source level and the sub-warehouse structure at the application level, as well as the hierarchical relationship between the sub-warehouse structure at the aggregation level and the sub-warehouse structure at the application level can all be represented by relationship lines.

[0083] like Figure 3As shown, the sub-warehouse structure at the source level includes table structure s1, table structure s2, table structure s3 and table structure s4; the sub-warehouse structure at the aggregation level includes table structure s5 and table structure s6; the sub-warehouse structure at the application level includes table structure s7; among them, table structure s5 is obtained by aggregation based on table structure s1, table structure s2 and table structure s3, table structure s6 is obtained by aggregation based on table structure s1 and table structure s4, and table structure s7 is determined based on table structure s1 and table structure s5, so there is a relationship line h1 between table structure s5 and table structure s1, table structure s2 and table structure s3; there is a relationship line h2 between table structure s6 and table structure s1 and table structure s4; there is a relationship line h3 between table structure s7 and table structure s1 and table structure s5.

[0084] In some embodiments, querying entity data of a business based on a first query message includes: replying to the first query message through a big model to obtain entity data of the business; replying to the second query message to obtain key entity data corresponding to the business demand information, including: replying to the second query message through a big model to obtain key entity data corresponding to the business demand information; executing a question-and-answer task based on a third query message to obtain table structures at different levels and hierarchical relationships between table structures, including: executing a question-and-answer task corresponding to the third query message through a big model to obtain table structures at different levels and hierarchical relationships between table structures.

[0085] Specifically, the method for generating data warehouse structure includes four stages, such as Figure 4 As shown, the first stage: generates the first query information based on the business knowledge data, inputs the first query information into the big model, and obtains the business entity data based on the response output by the big model to the first query information; the second stage: generates the second query information based on the business entity data and business demand information, inputs the second query information into the big model, and obtains the key entity data based on the response output by the big model to the second query information; the third stage: generates the third query information based on the key entity data, business demand information and entity constraint information, inputs the third query information into the big model, and obtains the table structures at different levels and the hierarchical relationships between the table structures based on the response output by the big model to the third query information; the fourth stage: constructs the data warehouse structure based on the table structures at different levels and the hierarchical relationships between the table structures.

[0086] In the above embodiment, through multiple intelligent question-and-answer tasks, the entity data of the business, the key entity data for realizing the functions corresponding to the business demand information, the table structures at different levels and the hierarchical relationships between the table structures are gradually determined, and then the data warehouse structure can be constructed based on the table structures at different levels and the hierarchical relationships between the table structures; compared with having the data warehouse structure constructed by experts in the business field, the data warehouse structure can be constructed automatically, which greatly improves the efficiency of constructing the data warehouse structure.

[0087] In some embodiments, as Figure 5 As shown, the method for generating a data warehouse structure also includes: step 501, generating first sample query information based on sample knowledge data, processing the first sample query information through the initial large model to obtain training entity data; step 502, generating second sample query information based on the training entity data and sample requirement information, processing the second sample query information through the initial large model to obtain training key entity data; step 503, generating third sample query information based on the training key entity data, sample requirement information and sample entity constraint information, processing the third sample query information through the initial large model to obtain training table structures at different levels and hierarchical relationships between the training table structures; step 504, constructing a training data warehouse structure based on training table structures at different levels and the hierarchical relationships between the training table structures; step 505, adjusting parameters of the initial large model based on the training data warehouse structure and the label data warehouse structure until the initial large model converges to obtain a large model.

[0088] Among them, the sample knowledge data is business-related knowledge data; the sample knowledge data used in training and the knowledge data used in application can correspond to the same business; the first sample query information is used to query the training entity data extracted based on the sample knowledge data; the initial large model and the large model have the same model structure.

[0089] The sample requirement information is used to represent the functions that the business needs to implement; the second sample query information is used to inquire about the training key entity data corresponding to the sample requirement information; the sample entity constraint information is used to reflect the entity's restriction conditions, specifically the restriction conditions of the entity's corresponding table structure; the third sample query information is used to inquire about the different levels of training table structures involved in implementing the functions corresponding to the sample requirement information and the hierarchical relationship between the training table structures.

[0090] The label data warehouse structure is a standard data warehouse structure under sample requirement information and sample entity constraint information; the label data warehouse structure includes sample table structures at different levels and hierarchical relationships between sample table structures at different levels.

[0091] Specifically, if Figure 6As shown, the parameters of the initial large model are adjusted according to the training data warehouse structure and the label data warehouse structure, which can be to obtain the training table structure belonging to the source level, the training table structure of the aggregation level and the training table structure of the application level in the training data warehouse structure; obtain the label table structure belonging to the source level, the label table structure of the aggregation level and the label table structure of the application level in the label data warehouse structure; calculate the first loss value according to the training table structure and the label table structure of the source level, calculate the second loss value according to the training table structure and the label table structure of the aggregation level, and calculate the third loss value according to the training table structure and the label table structure of the application level; determine the first fusion loss value according to the first loss value, the second loss value and the third loss value, and adjust the parameters of the initial large model based on the first fusion loss value until the initial large model converges to obtain the large model.

[0092] In some embodiments, training the initial large model may include a process of improving the initial large model's ability to extract entity data of the business, extract key entity data, and determine table structures at different levels and hierarchical relationships between table structures.

[0093] like Figure 7 As shown, the process of training the large model may also include: generating first sample query information based on sample knowledge data, processing the first sample query information through the initial large model to obtain training entity data; determining a fourth loss value based on the training entity data and the entity data label; generating second sample query information based on the training entity data and sample requirement information, processing the second sample query information through the initial large model to obtain training key entity data, and determining a fifth loss value based on the training key entity data and the key entity data label; generating third sample query information based on the training key entity data, sample requirement information and sample entity constraint information, processing the third sample query information through the initial large model to obtain training table structures at different levels and hierarchical relationships between training table structures, constructing a training data warehouse structure based on the training table structures at different levels and the hierarchical relationships between training table structures, and determining a sixth loss value based on the training data warehouse structure and the label data warehouse structure; determining a second fusion loss value based on the fourth loss value, the fifth loss value and the sixth loss value, and adjusting the parameters of the initial large model based on the second fusion loss value until the initial large model converges to obtain the large model.

[0094] The sixth loss value may be the same as the first fusion loss value in the above embodiment.

[0095] In the above embodiment, the initial big model is trained so that the trained big model has the ability to extract business entity data, the ability to extract key entity data, and the ability to determine the hierarchical relationship between table structures at different levels. Then, the big model can build a data warehouse structure by executing multiple question-and-answer tasks, thereby improving the efficiency of building a data warehouse structure.

[0096] In the above-mentioned method for generating a data warehouse structure, through multiple intelligent question-answering tasks, the entity data of the business is gradually determined, the key entity data of the functions corresponding to the business requirement information is extracted, and the table structures at different levels that meet the business requirement information and entity constraint information and the hierarchical relationships between the table structures are determined. Then, the data warehouse structure can be constructed based on the table structures at different levels and the hierarchical relationships between the table structures. Compared with the manual construction of the data warehouse structure corresponding to the business requirement information by experts in the business field, the data warehouse structure corresponding to the business requirement information can be automatically constructed, thereby improving the efficiency of constructing the data warehouse structure.

[0097] In some embodiments, as Figure 8 As shown, the method for generating a data warehouse structure also includes: step 801, generating a fourth query information for performing data warehouse verification based on the data warehouse structure to be processed; step 802, replying to the fourth query information through the large model to obtain the abnormal table structure in the data warehouse structure to be processed.

[0098] The data warehouse structure to be processed is the data warehouse to be verified; the fourth inquiry information is used to inquire about abnormal conditions of the data warehouse structure to be processed; and the abnormal table structure includes at least one of an erroneous table structure and a redundant table structure.

[0099] Specifically, the computer device obtains the structure of the data warehouse to be processed, converts the structure of the data warehouse to be processed into a data dictionary, embeds the data dictionary corresponding to the structure of the data warehouse to be processed into a template for inquiring about abnormal situations in the data warehouse, and obtains fourth query information; inputs the fourth query information into the big model, and obtains the abnormal table structure in the structure of the data warehouse to be processed through the reply content output by the big model; in actual applications, the abnormal table structure in the structure of the data warehouse to be processed can be represented by the primary key of the abnormal table structure.

[0100] In one implementation, the computer device may also convert the warehouse structure of the data to be processed into an ER diagram (entity relationship diagram), and determine the fourth query information according to the ER diagram corresponding to the warehouse structure of the data to be processed.

[0101] In some embodiments, after determining the abnormal table structure in the data warehouse structure to be processed, an improvement report can be output based on the abnormal table structure. For example, if the abnormal table structure is a redundant table structure, the improvement report can be: there is redundant data in the abnormal table structure, and it is recommended to merge the abnormal table structure.

[0102] In some embodiments, after determining the abnormal table structure in the data warehouse structure to be processed, the data warehouse structure to be processed can be improved based on the abnormal table structure to obtain the target data warehouse structure; specifically, when the abnormal table structure is a redundant table structure, the abnormal table structure is merged; when the abnormal table structure is an incorrect table structure, the abnormal table structure is deleted or modified; based on the abnormal table structure, the data warehouse structure to be processed is improved to obtain the target data warehouse structure, so that the target data warehouse structure is concise and correct, and business is processed based on the target data warehouse structure, thereby improving business processing efficiency.

[0103] In the above embodiment, the data warehouse structure to be processed is verified by a large model to determine the abnormal table structure in the data warehouse structure to be processed, so that the abnormal table structure in the data warehouse structure to be processed can be improved subsequently, providing a reference for the improvement of the data warehouse structure to be processed, thereby improving the business processing efficiency of the data warehouse structure.

[0104] In some embodiments, as Figure 9 As shown, the method for generating a data warehouse structure also includes: step 901, modifying the label data warehouse structure to obtain an abnormal data warehouse structure; step 902, generating a fourth sample query information based on the abnormal data warehouse structure; step 903, replying to the fourth sample query information through the large model to obtain a training abnormality table structure; step 904, determining the label abnormality table structure based on the difference between the label data warehouse structure and the abnormal data warehouse structure; step 905, adjusting the parameters of the large model according to the training abnormality table structure and the label abnormality table structure until the parameters of the large model converge to obtain an updated large model.

[0105] Among them, the label data warehouse structure is a data warehouse structure without redundant or erroneous table structures, and the label data warehouse can be determined based on excellent data warehouse design cases; the abnormal data warehouse structure is a data warehouse structure with redundant or erroneous table structures; the fourth sample query information is used to inquire about abnormal situations of the abnormal data warehouse structure; the training abnormal table structure is the result obtained by the large model to verify the abnormal data warehouse structure; the label abnormal table structure is determined based on the difference between the label data warehouse structure and the abnormal data warehouse structure.

[0106] Specifically, the computer device obtains the label data warehouse structure, modifies the label data warehouse structure to obtain the exception data warehouse structure, compares the differences between the label data warehouse structure and the exception data warehouse structure, and obtains the label exception table structure; generates a fourth sample query information according to the exception data warehouse structure, inputs the fourth sample query information into the large model, obtains the training exception table structure through the response output by the large model, calculates the loss value according to the training exception table structure and the label exception table structure, adjusts the parameters of the large model according to the loss value, until the parameters of the large model converge, and obtains the updated large model.

[0107] The loss value is calculated according to the training anomaly table structure and the label anomaly table structure. The methods include obtaining each training attribute of the training anomaly table structure, obtaining each label attribute of the label anomaly table structure, extracting the first eigenvector of each training attribute, extracting the second eigenvector of each label attribute, performing feature fusion on each extracted first eigenvector to obtain a first training feature vector, and performing feature fusion on each extracted second eigenvector to obtain a second training feature vector; and taking the similarity between the first training feature vector and the second training feature vector as the loss value.

[0108] Performing feature fusion on each extracted first eigenvector to obtain a first training eigenvector can be performed by determining the fusion weight of each first eigenvector according to the training attribute corresponding to each first eigenvector, and performing feature fusion on each first eigenvector according to the fusion weight of each first eigenvector to obtain the first training eigenvector; determining the fusion weight of each first eigenvector according to the training attribute corresponding to each first eigenvector, which can be a fusion weight corresponding to a training attribute as a primary key, which is greater than a fusion weight corresponding to a training attribute as a foreign key.

[0109] Similarly, performing feature fusion on the extracted second feature vectors to obtain the second training feature vector can also be achieved in the same manner as performing feature fusion on the first feature vectors.

[0110] It should be noted that the updated large model can replace the large model before the update and be used to perform the question-answering tasks corresponding to the first query information, the second query information, and the third query information.

[0111] In the above embodiment, the abnormal data warehouse structure is obtained by modifying the label data warehouse structure, rather than obtaining the abnormal data warehouse structure with an abnormal table structure, which reduces the difficulty of obtaining abnormal samples and improves the efficiency of obtaining abnormal samples; the fourth sample query information is processed by the large model, and the large model is trained, so that the large model learns the abnormal information in the abnormal data warehouse structure, and then the large model can verify the data warehouse structure to be processed and obtain the abnormal table structure in the data warehouse structure to be processed, which provides a reference for the improvement of the data warehouse structure to be processed, and thus can improve the business processing efficiency of the data warehouse structure.

[0112] In some embodiments, the second query information is replied to and processed to obtain key entity data corresponding to the business requirement information, including: functionally splitting the business requirement information included in the second query information to obtain each sub-requirement information; and selecting key entity data corresponding to each sub-requirement information from the entity data included in the second query information.

[0113] The business requirement information can be used to implement the business function, and the sub-requirement information is used to implement the subdivision of the function.

[0114] Exemplarily, the business demand information includes: querying the annual income of the user in a specified year, and performing functional decomposition of the business demand information. The sub-demand information obtained includes: querying each income and expenditure data of the user in the specified year (sub-demand information 1), filtering income data from the income and expenditure data (sub-demand information 2), and adding the filtered income data to obtain the annual income of the specified year (sub-demand information 3).

[0115] Specifically, the computer device obtains the business requirement information included in the second inquiry information, splits the functions implemented by the business requirement information, and obtains sub-requirement information corresponding to each sub-function under the implemented function; for each sub-requirement information, the key entity data corresponding to the sub-requirement information is selected from the entity data.

[0116] For example, sub-demand information 1 is: querying each income and expenditure data of the user in a specified year. The key entity data corresponding to sub-demand information 1 includes user entity data, year entity data, and income and expenditure entity data; sub-demand information 2 is: filtering income data from income and expenditure data. The key entity data corresponding to sub-demand information 2 includes: income entity data; the key entity data corresponding to sub-demand information 3 includes: annual income entity data.

[0117] It should be noted that the above process of replying to the second inquiry information and obtaining the key entity data corresponding to the business requirement information can be implemented through a large model.

[0118] In the above embodiment, based on the functions realized by the business requirement information, the business requirement information is refined to obtain sub-requirement information, and key entity data is determined based on the sub-requirement information, thereby improving the accuracy of the key entity data.

[0119] In some embodiments, a question-and-answer task is performed based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures, including: selecting key entity data at different levels based on business requirement information from the key entity data included in the third query information, and determining the dependency relationships between key entity data at different levels; determining the hierarchical relationships between table structures at different levels based on the key entity data at different levels, the dependency relationships, and the entity constraint information included in the third query information.

[0120] Among them, key entity data at different levels include: key entity data at the source level, key entity data at the aggregation level, and key entity data at the application level; the key entity data at the source level is the relevant data of the key entities at the source level, the key entity data at the aggregation level is the relevant data of the key entities at the aggregation level, and the key entity data at the application level is the relevant data of the key entities at the application level.

[0121] Specifically, the computer device can functionally decompose the business demand information to obtain sub-demand information, and each sub-demand information can constitute a complete business logic. The entities involved in each sub-demand information can be hierarchically divided to obtain key entities at different levels. Based on the key entities at different levels, key entity data at different levels can be obtained from the key entity data included in the third query information. Since each sub-demand information can constitute a complete business logic, the dependency relationship between each key entity can be determined according to the business logic, and then the dependency relationship between each key entity data can be determined.

[0122] For example, if it is determined according to business logic that key entity k1 needs to be obtained based on key entity k2, there is a dependency relationship between key entity k1 and key entity k2, and further a dependency relationship exists between key entity data of key entity k1 and key entity data of key entity k2.

[0123] After determining the key entity data at different levels and the dependency relationships between the key entity data at different levels, the table structures of the key entities at different levels are constructed based on the key entity data and entity constraint information at different levels. Based on the dependency relationships between the key entity data at different levels, the hierarchical relationships between the table structures at different levels (corresponding to the key entity data) are determined.

[0124] It should be noted that the table structure of the key entity determined based on the key entity data can be used to reflect the name, entity attributes and entity relationship of the key entity; in actual applications, the table structure can be a table that does not include data, or it can be an SQL statement corresponding to the table structure.

[0125] It should be noted that the above process of performing the question-answering task based on the third query information and obtaining the table structures at different levels and the hierarchical relationships between the table structures can be implemented through a large model.

[0126] In the above embodiment, key entity data at different levels are selected based on business demand information, and the dependency relationships between key entity data at different levels are determined, thereby obtaining table structures at different levels and hierarchical relationships between table structures. This makes the hierarchical relationships between the levels to which the table structures belong and the table structures strongly correlated with the business demand information, thereby improving the accuracy of the obtained hierarchical relationships between table structures at different levels and the table structures.

[0127] In some embodiments, among the key entity data included in the third query information, selecting key entity data at different levels based on business requirement information and determining dependencies between the key entity data at different levels include:

[0128] Determine logical requirement information and display requirement information based on the business requirement information; select, from the key entity data included in the third query information, first key entity data belonging to the source level and second key entity data belonging to the aggregation level based on the logical requirement information, and determine a first dependency relationship between the first key entity data and the second key entity data; select, from the key entity data, third key entity data belonging to the application level based on the display requirement information, and determine a second dependency relationship among the first key entity data, the second key entity data, and the third key entity data.

[0129] Among them, logical requirement information is used to represent the underlying functions that the business needs to implement (corresponding to invisible data), and display requirement information is used to represent the user-oriented interactive functions that the business needs to implement (corresponding to visible data). For example, the underlying function of the business is to count the order data of all regions, and the interactive function of the business is to display the annual income queried by the user.

[0130] The number of each of the first key entity data, the second key entity data, and the third key entity data may be at least one.

[0131] The first dependency relationship between the first key entity data and the second key entity data indicates that the second key entity needs to be determined by relying on the first key entity; the second dependency relationship between the first key entity data, the second key entity data and the third key entity data indicates that the third key entity needs to be determined by relying on at least one of the first key entity or the second key entity.

[0132] Specifically, the computer device performs functional decomposition on the business requirement information to obtain sub-requirement information, and determines the logical requirement information and display requirement information in each sub-requirement information; obtains each key entity involved in the logical requirement information, and divides the key entities involved in the logical requirement information into levels to obtain a first key entity of the source level and a second key entity of the aggregation level that implement the corresponding function of the logical requirement information, and determines a first dependency relationship between the first key entity of the source level and the second key entity of the aggregation level based on the logical requirement information, and then selects the first key entity data belonging to the source level and the second key entity data belonging to the aggregation level from the key entity data, and obtains a first dependency relationship between the first key entity data and the second key entity data.

[0133] Key entities involved in the display demand information are obtained, and the key entities involved in the display demand information are divided into levels to obtain a first key entity at the source level, a second key entity at the aggregation level, and a third key entity at the application level that implement the function corresponding to the display demand information. Based on the display demand information, a second dependency relationship among the first key entity at the source level, the second key entity at the aggregation level, and the third key entity at the application level is determined. Then, from the key entity data, the first key entity data belonging to the source level, the second key entity data belonging to the aggregation level, and the third key entity data belonging to the application level are selected, and a second dependency relationship among the first key entity data, the second key entity data, and the third key entity data is obtained.

[0134] It should be noted that the first key entity data and the second key entity data determined according to the display requirement information may be the same as the first key entity data and the second key entity data determined according to the logic requirement information.

[0135] In actual applications, the third key entity at the application level may depend on the first key entity at the source level, or may depend on the second key entity at the aggregation level, or may depend on the first key entity at the source level and the second key entity at the aggregation level. Therefore, based on the display requirement information, it can be determined whether it depends on at least one of the first key entity at the source level or the second key entity at the aggregation level.

[0136] Exemplarily, the business demand information includes: querying the changes in the user's annual income within a specified time range, the logical demand information determined based on the business demand information includes: obtaining the user's income and expenditure data within the specified time range, and determining the user's annual income within the specified time range based on the income and expenditure data, and the display demand information includes: displaying the changes in annual income.

[0137] like Figure 10As shown, the first key entity of the source level determined according to the logical demand information includes: user entity, time entity and income and expenditure entity, and the second key entity of the aggregation level determined according to the logical demand information includes: user's annual income entity; the first key entity of the source level determined according to the display demand information includes: time entity; the second key entity of the aggregation level determined according to the display demand information includes: user's annual income entity, and the third key entity of the application level determined according to the display demand information includes: user's annual income change entity within a specified time range.

[0138] The first dependency relationship includes: the dependency relationship between the user entity and income and expenditure entity at the source level and the user's annual income entity at the aggregation level. The second dependency relationship includes: the dependency relationship between the time entity at the source level, the user's annual income entity at the aggregation level and the third key entity at the application level (the user's annual income change entity within a specified time range).

[0139] Based on the key entities of each level determined above, and the first dependency relationship and the second dependency relationship between the key entities of each level, the first key entity data belonging to the source level, the second key entity data belonging to the aggregation level, and the third key entity data belonging to the application level are selected from the key entity data, and the first dependency relationship between the first key entity data and the second key entity data is determined, and the second dependency relationship between the first key entity data, the second key entity data, and the third key entity data is determined.

[0140] It should be noted that the relationship between the key entity data at each level and the key entity data shown in the above examples is relatively simple. In actual applications, complex business demand information may involve multiple key entities at different levels, and the dependency relationship between multiple key entities is relatively complex. For example, the aggregation level can also include multiple sub-aggregation levels to realize multi-level aggregation operations between entity corresponding table structures.

[0141] It should be noted that the process of selecting key entity data at different levels from the key entity data included in the third query information according to business requirement information and determining the dependency relationships between the key entity data at different levels can be implemented through a large model.

[0142] In the above embodiment, logical requirement information and display requirement information are determined based on business requirement information. Among the key entity data, the key entity data of the source level, aggregation level and application level are determined according to the logical requirement information and display requirement information, and the dependency relationship between the key entity data of each level is determined, so that the key entity data and the dependency relationship are related to the business requirement information, which improves the matching degree between the key entity data of different levels and the business requirement information, and thus can improve the accuracy of the subsequently determined table structures of different levels and the hierarchical relationship between the table structures. Moreover, this process can be executed automatically without manual participation, which improves the efficiency of determining the hierarchical relationship between the table structures of different levels and the table structures.

[0143] In some embodiments, the table structure includes a first table structure, a second table structure, and a third table structure; the hierarchical relationship includes a first hierarchical relationship and a second hierarchical relationship; based on the key entity data, dependency relationships, and entity constraint information included in the third query information at different levels, the hierarchical relationship between the table structures at different levels is determined, including: based on the first key entity data, the second key entity data, the first dependency relationship, and the entity constraint information included in the third query information, a first table structure belonging to the source level and a second table structure belonging to the aggregation level are constructed; based on the first dependency relationship, a first hierarchical relationship between the first table structure and the second table structure is determined; based on the first key entity data, the second key entity data, the third key entity data, the second dependency relationship, and the entity constraint information, a third table structure belonging to the application level is constructed; based on the second dependency relationship, a second hierarchical relationship between the first table structure, the second table structure, and the third table structure is determined.

[0144] Specifically, the computer device can construct a first table structure belonging to the source level based on the first key entity data, construct a second table structure belonging to the aggregation level based on the second key entity data, and construct a third table structure belonging to the application level based on the third key entity data; determine the first-level relationship between the first table structure corresponding to the first key entity data and the second table structure corresponding to the second key entity data based on the first dependency relationship between the first key entity data and the second key entity data; determine the second-level relationship between the first table structure corresponding to the first key entity data, the second table structure corresponding to the second key entity data, and the third table structure corresponding to the third key entity data based on the second dependency relationship between the first key entity data, the second key entity data, and the third key entity data.

[0145] In practical applications, the first-level relationship between the first table structure and the second table structure can be described by text. For example, there is a hierarchical relationship between the first table structure f1 and the second table structure f2. Similarly, the second-level relationship between the first table structure, the second table structure, and the third table structure can also be described by text.

[0146] It should be noted that the above process of determining the table structures at different levels and the hierarchical relationships between the table structures based on the key entity data at different levels, dependency relationships and entity constraint information included in the third query information can be implemented through a large model.

[0147] In the above embodiment, based on the key entity data at different levels and the dependency relationships between the key entity data, the corresponding table structures of the key entity data at different levels and the hierarchical relationships between the table structures are determined. This process does not require manual participation, thereby improving the efficiency of determining the table structures at different levels and the hierarchical relationships between the table structures.

[0148] In some embodiments, the table structure includes a first table structure, a second table structure, and a third table structure; performing a question-and-answer task based on the third query information to obtain table structures at different levels and hierarchical relationships between the table structures includes: determining logical requirement information and display requirement information based on the business requirement information;

[0149] Among the key entity data included in the third query information, the first key entity data of the source level is selected based on the logical requirement information; based on the first key entity data and the entity constraint information included in the third query information, a first table structure belonging to the source level is constructed; based on the first table structure, the logical requirement information and the display requirement information, a second table structure belonging to the aggregation level and a third table structure belonging to the application level are constructed, and the hierarchical relationship among the first table structure, the second table structure and the third table structure is determined.

[0150] Among them, the specific process of determining logical requirement information and displaying requirement information based on business requirement information can refer to the description in the above embodiment; the computer device obtains the key entities involved in the logical requirement information, determines the first key entity belonging to the source level among the key entities involved in the logical requirement information, determines the first key entity data belonging to the source level among the key entity data included in the third query information, and constructs the first table structure belonging to the source level based on the first key entity data of the source level.

[0151] In some embodiments, among the entities involved in the logical requirement information, each second key entity that has a dependency relationship with the first key entity is determined, and second key entity data corresponding to the second key entity is obtained from the key entity data included in the third query information, and a second table structure belonging to the aggregation level is constructed based on the second key entity data.

[0152] Among the entities involved in the display demand information, determine the candidate key entities that have a dependency relationship with the first key entity. For each candidate key entity, the candidate key entity can be the second key entity. If the candidate key entity is not the second key entity, then the candidate key entity is used as the third key entity; obtain the third key entity data corresponding to the third key entity from the key entity data included in the third query information, and construct a third table structure belonging to the application level based on the third key entity data.

[0153] According to the dependency relationship between the first key entity and the second key entity, and the dependency relationship between the first key entity and the candidate key entity (the second key entity or the third key entity), a hierarchical relationship among the first table structure, the second table structure, and the third table structure is determined.

[0154] It should be noted that the above process of performing the question-answering task based on the third query information and obtaining the table structures at different levels and the hierarchical relationships between the table structures can be implemented through a large model.

[0155] In the above embodiment, based on the business demand information, the hierarchical relationship between the table structures at different levels and the table structures is determined, so that the level to which the table structure belongs and the hierarchical relationship between the table structures are strongly correlated with the business demand information, thereby improving the accuracy of the obtained hierarchical relationship between the table structures at different levels and the table structures.

[0156] In some embodiments, the hierarchical relationship includes a first hierarchical relationship and a second hierarchical relationship; constructing a second table structure belonging to an aggregation level and a third table structure belonging to an application level based on the first table structure, the logic requirement information, and the display requirement information, and determining the hierarchical relationship among the first table structure, the second table structure, and the third table structure, including:

[0157] A first candidate table structure is selected from the first table structure based on the logical requirement information; the first candidate table structure is aggregated to obtain a second table structure belonging to the aggregation level; a second candidate table structure is selected from the first table structure and the second table structure based on the display requirement information; the second candidate table structure is extracted to obtain a third table structure belonging to the application level; a first-level relationship between the first candidate table structure and the second table structure is determined, and a second-level relationship between the second candidate table structure and the third table structure is determined.

[0158] Specifically, the logical requirement information includes at least one logical requirement sub-information arranged in the execution logic order. For each logical requirement sub-information, the first candidate structure table related to the logical requirement sub-information is obtained in the first table structure belonging to the source level, and the first candidate structure table is aggregated to obtain the second structure table corresponding to the logical requirement sub-information and belonging to the aggregation level.

[0159] The display demand information includes at least one display demand sub-information. For each display demand sub-information, the second candidate structure table involved in the display demand sub-information is obtained in the first table structure belonging to the source level and the second table structure belonging to the aggregation level, and data extraction processing is performed on the second candidate structure table to obtain the third table structure belonging to the application level. It should be noted that since the third table structure of the application level is visible data for user interaction, this visible data can be extracted from the table structures of the source level and the aggregation level, or extracted from the table structures of the source level and the aggregation level and then subjected to simple data processing. Therefore, data extraction processing is performed on the second candidate structure to obtain the third table structure belonging to the application level.

[0160] Since the second table structure is determined based on the first candidate table structure, there is a dependency relationship between the first candidate table structure and the second table structure. Based on this dependency relationship, the first-level relationship between the first candidate table structure and the second table structure can be determined; similarly, since the third table structure is determined based on the second candidate table structure, there is a dependency relationship between the third table structure and the second candidate table structure. Based on this dependency relationship, the second-level relationship between the second candidate table structure and the third table structure can be determined.

[0161] It should be noted that the above process of constructing the second table structure belonging to the aggregation level and the third table structure belonging to the application level based on the first table structure, logical requirement information and display requirement information, and determining the hierarchical relationship between the first table structure, the second table structure and the third table structure can be achieved through a large model.

[0162] In the above embodiment, based on the logical requirement information and the display requirement information, the first table structure belonging to the source level is aggregated to obtain the second table structure belonging to the aggregation level, the first table structure and the second table structure are processed to obtain the third table structure belonging to the application level, and the hierarchical relationship between the table structures is determined, so that the determined table structures at different levels and the hierarchical relationship between the table structures are strongly correlated with the business requirement information, thereby improving the accuracy of the table structures at different levels and the hierarchical relationship between the table structures.

[0163] In some embodiments, the data request processing method can be used to construct a data warehouse structure for transaction services and verify the data warehouse structure to be processed for transaction services.

[0164] The computer device generates first query information through the knowledge data of the transaction business, processes the first query information through the big model, and obtains the entity data of the transaction business; illustratively, the entity data includes: the entity data of the user, transaction data, transaction medium card, transaction object, registered user of the transaction object, and transaction area.

[0165] The second query information is generated through the entity data of the transaction business and the business demand information of the transaction business. The second query information is processed through the large model to obtain the key entity data corresponding to the business demand information. For example, the business demand information includes: querying the transaction data of the transaction object in area A; the key entity data includes: the entity data of the user, transaction data, transaction object and transaction area.

[0166] The third query information is generated through key entity data, business requirement information and entity constraint information, and the third query information is processed through the large model to obtain table structures at different levels and the hierarchical relationships between table structures.

[0167] Exemplarily, entity constraint information may be a restriction condition of a table structure, and table structures at different levels include: belonging to the source level: transaction object table structure, transaction data table structure, and transaction area table structure; belonging to the aggregation level: transaction data table structure of the transaction object and transaction data table structure of area A; belonging to the application level: transaction data table structure of the transaction object in area A; wherein, the transaction object table structure and the transaction data table structure have a hierarchical relationship with the transaction data table structure of the transaction object, and the transaction data table structure and the transaction area table structure have a hierarchical relationship with the transaction data table structure of area A; the transaction data table structure of the transaction object and the transaction data table structure of area A have a hierarchical relationship with the transaction data table structure of the transaction object in area A; Figure 11 As shown, according to the table structures at different levels and the hierarchical relationships between the table structures, the data warehouse structure of the transaction business is constructed; the hierarchical relationships between the table structures at different levels are represented by dotted lines, for example, according to Figure 11 As can be seen from the dotted line, the transaction object table structure and transaction data table structure at the source level have a hierarchical relationship with the transaction data table structure of the transaction object at the aggregation level.

[0168] In some embodiments, the data request processing method can be used to verify and improve the warehouse structure of data to be processed for financial services.

[0169] Obtain the warehouse structure of the data to be processed under the financial business, generate the fourth query information according to the warehouse structure of the data to be processed, process the fourth query information through the large model, and obtain the abnormal table structure in the warehouse structure of the data to be processed; for example, Figure 12As shown, the data warehouse structure to be processed under the financial business includes the following: user table structure, regional table structure, rate table structure and account table structure belonging to the source level; and the account table structure of the user in region A belonging to the aggregation level; and there is a hierarchical relationship between the user table structure, regional table structure, rate table structure and account table structure and the account table structure of the user in region A; the fourth query information corresponding to the data warehouse structure to be processed is processed by the big model, and the abnormal table structure includes: regional table structure and rate table structure. It can be seen that there is a large amount of redundant data in the regional table structure and the rate table structure. Subsequently, the regional table structure and the rate table structure can be merged into the regional rate table structure, so that the improved data warehouse structure is simpler. For example, in the improved data warehouse structure, the user table structure, regional rate table structure and account table structure belong to the source level; and there is a hierarchical relationship between the user table structure, regional rate table structure and account table structure and the account table structure of the user in region A.

[0170] In some embodiments, as Figure 13 As shown, the method for generating the data warehouse structure includes:

[0171] Step 1301: Generate first sample query information based on sample knowledge data, process the first sample query information using the initial large model to obtain training entity data; generate second sample query information based on the training entity data and sample requirement information, process the second sample query information using the initial large model to obtain training key entity data; generate third sample query information based on the training key entity data, sample requirement information, and sample entity constraint information, process the third sample query information using the initial large model to obtain training table structures at different levels and hierarchical relationships between the training table structures; construct a training data warehouse structure based on the training table structures at different levels and the hierarchical relationships between the training table structures; adjust parameters of the initial large model based on the training data warehouse structure and the label data warehouse structure until the initial large model converges to obtain a large model;

[0172] Step 1302: Modify the label data warehouse structure to obtain an anomaly data warehouse structure; generate a fourth sample query message based on the anomaly data warehouse structure; use the large model to process the fourth sample query message to obtain a training anomaly table structure; determine the label anomaly table structure based on the difference between the label data warehouse structure and the anomaly data warehouse structure; adjust the parameters of the large model based on the training anomaly table structure and the label anomaly table structure until the parameters of the large model converge, thereby obtaining an updated large model;

[0173] Step 1303: Generate first query information based on the business knowledge data, and respond to the first query information using the big model to obtain entity data of the business.

[0174] Step 13041: Generate second query information based on the business entity data and business requirement information, and respond to the second query information through the big model to obtain key entity data corresponding to the business requirement information;

[0175] Step 13051: Generate third query information based on key entity data, business requirement information, and entity constraint information. Execute the question-answering task corresponding to the third query information through the big model to obtain table structures at different levels and the hierarchical relationships between the table structures.

[0176] Step 13042: Generate second query information based on the business entity data and business requirement information, perform functional decomposition on the business requirement information included in the second query information to obtain sub-requirement information; and select key entity data corresponding to each sub-requirement information from the entity data included in the second query information.

[0177] Step 130521: Generate third query information based on the key entity data, business requirement information, and entity constraint information, and determine logical requirement information and presentation requirement information based on the business requirement information. From the key entity data included in the third query information, select first key entity data belonging to the source level and second key entity data belonging to the aggregation level based on the logical requirement information, and determine a first dependency relationship between the first key entity data and the second key entity data. From the key entity data, select third key entity data belonging to the application level based on the presentation requirement information, and determine a second dependency relationship between the first key entity data, the second key entity data, and the third key entity data.

[0178] Step 130522: Based on the first key entity data, the second key entity data, the first dependency relationship, and the entity constraint information included in the third query information, a first table structure belonging to the source level and a second table structure belonging to the aggregation level are constructed; based on the first dependency relationship, a first-level relationship between the first table structure and the second table structure is determined; based on the first key entity data, the second key entity data, the third key entity data, the second dependency relationship, and the entity constraint information, a third table structure belonging to the application level is constructed; based on the second dependency relationship, a second-level relationship between the first table structure, the second table structure, and the third table structure is determined; the table structure includes the first table structure, the second table structure, and the third table structure; and the hierarchical relationship includes the first-level relationship and the second-level relationship.

[0179] Step 130531: Generate third query information based on the key entity data, business requirement information, and entity constraint information. Determine logical requirement information and display requirement information based on the business requirement information. Select the first key entity data at the source level from the key entity data included in the third query information based on the logical requirement information. Construct a first table structure at the source level based on the first key entity data and the entity constraint information included in the third query information.

[0180] Step 130532: Select a first candidate table structure from the first table structure based on the logical requirement information; perform aggregation processing on the first candidate table structure to obtain a second table structure belonging to the aggregation level; select a second candidate table structure from the first and second table structures based on the display requirement information; perform data extraction processing on the second candidate table structure to obtain a third table structure belonging to the application level; determine a first-level relationship between the first candidate table structure and the second table structure, and determine a second-level relationship between the second candidate table structure and the third table structure; the table structure includes the first table structure, the second table structure, and the third table structure; and the hierarchical relationship includes a first-level relationship and a second-level relationship.

[0181] Step 1306: construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between the table structures;

[0182] Step 1307: Generate fourth query information for data warehouse verification based on the data warehouse structure to be processed; reply to the fourth query information through the large model to obtain the abnormal table structure in the data warehouse structure to be processed.

[0183] In the above-mentioned method for generating a data warehouse structure, through multiple intelligent question-answering tasks, the entity data of the business is gradually determined, the key entity data of the functions corresponding to the business requirement information is extracted, and the table structures at different levels that meet the business requirement information and entity constraint information and the hierarchical relationships between the table structures are determined. Then, the data warehouse structure can be constructed based on the table structures at different levels and the hierarchical relationships between the table structures. Compared with the manual construction of the data warehouse structure corresponding to the business requirement information by experts in the business field, the data warehouse structure corresponding to the business requirement information can be automatically constructed, thereby improving the efficiency of constructing the data warehouse structure.

[0184] It should be understood that, although the steps in the flowcharts of the above embodiments are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the flowcharts of the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily to be performed in sequence, but can be performed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0185] Based on the same inventive concept, embodiments of the present application also provide a device for generating a data warehouse structure for implementing the aforementioned method for generating a data warehouse structure. The implementation solution provided by this device is similar to the implementation solution described in the aforementioned method. Therefore, the specific limitations of the one or more embodiments of the device for generating a data warehouse structure provided below can be found in the aforementioned limitations of the method for generating a data warehouse structure, and will not be repeated here.

[0186] In some embodiments, as Figure 14 As shown, a device for generating a data warehouse structure is provided, comprising: an entity data determination module 1401, a key entity data determination module 1402, a table structure and hierarchical relationship determination module 1403, and a data warehouse structure construction module 1404, wherein:

[0187] The entity data determination module 1401 is configured to generate first query information based on the knowledge data of the business, and query the entity data of the business based on the first query information;

[0188] The key entity data determination module 1402 is configured to generate a second query message based on the business entity data and the business requirement information, and perform reply processing on the second query message to obtain the key entity data corresponding to the business requirement information;

[0189] Table structure and hierarchical relationship determination module 1403, configured to generate third query information based on key entity data, business requirement information, and entity constraint information, and perform a question-answering task based on the third query information to obtain table structures at different levels and the hierarchical relationships between the table structures;

[0190] The data warehouse structure construction module 1404 is used to construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between the table structures.

[0191] In some embodiments, the key entity data determination module 1402 is further configured to functionally decompose the business requirement information included in the second query information to obtain sub-requirement information; and select key entity data corresponding to each sub-requirement information from the entity data included in the second query information.

[0192] In some embodiments, the table structure and hierarchical relationship determination module 1403 is also used to select key entity data of different levels from the key entity data included in the third query information based on business requirement information, and determine the dependency relationships between the key entity data of different levels; based on the key entity data of different levels, the dependency relationships and the entity constraint information included in the third query information, determine the table structures of different levels and the hierarchical relationships between the table structures.

[0193] In some embodiments, the table structure and hierarchical relationship determination module 1403 is further used to determine logical requirement information and display requirement information based on business requirement information; among the key entity data included in the third query information, the first key entity data belonging to the source level and the second key entity data belonging to the aggregation level are selected based on the logical requirement information, and a first dependency relationship between the first key entity data and the second key entity data is determined; among the key entity data, the third key entity data belonging to the application level is selected based on the display requirement information, and a second dependency relationship between the first key entity data, the second key entity data, and the third key entity data is determined.

[0194] In some embodiments, the table structure includes a first table structure, a second table structure, and a third table structure; the hierarchical relationship includes a first hierarchical relationship and a second hierarchical relationship;

[0195] The table structure and hierarchical relationship determination module 1403 is further used to construct a first table structure belonging to the source level and a second table structure belonging to the aggregation level based on the first key entity data, the second key entity data, the first dependency relationship and the entity constraint information included in the third query information; determine the first hierarchical relationship between the first table structure and the second table structure based on the first dependency relationship; construct a third table structure belonging to the application level based on the first key entity data, the second key entity data, the third key entity data, the second dependency relationship and the entity constraint information; and determine the second hierarchical relationship between the first table structure, the second table structure and the third table structure based on the second dependency relationship.

[0196] In some embodiments, the table structure includes a first table structure, a second table structure, and a third table structure; the table structure and hierarchical relationship determination module 1403 is also used to determine logical requirement information and display requirement information based on business requirement information; among the key entity data included in the third query information, the first key entity data of the source level is selected according to the logical requirement information; based on the first key entity data and the entity constraint information included in the third query information, a first table structure belonging to the source level is constructed; based on the first table structure, the logical requirement information, and the display requirement information, a second table structure belonging to the aggregation level and a third table structure belonging to the application level are constructed, and the hierarchical relationship between the first table structure, the second table structure, and the third table structure is determined.

[0197] In some embodiments, the hierarchical relationship includes a first hierarchical relationship and a second hierarchical relationship; the table structure and hierarchical relationship determination module 1403 is further used to select a first candidate table structure in the first table structure based on the logical requirement information; perform aggregation processing on the first candidate table structure to obtain a second table structure belonging to the aggregation level; select a second candidate table structure from the first table structure and the second table structure based on the display requirement information; perform data extraction processing on the second candidate table structure to obtain a third table structure belonging to the application level; determine the first hierarchical relationship between the first candidate table structure and the second table structure, and determine the second hierarchical relationship between the second candidate table structure and the third table structure.

[0198] In some embodiments, the entity data determination module 1401 is further configured to process the first query information in response to the big model to obtain entity data of the business;

[0199] The key entity data determination module 1401 is further configured to respond to the second query information using the large model to obtain key entity data corresponding to the business requirement information;

[0200] The table structure and hierarchical relationship determination module 1403 is further configured to execute the question-answering task corresponding to the third query information through the large model to obtain table structures at different levels and the hierarchical relationships between the table structures.

[0201] In some embodiments, the data warehouse structure generation device also includes: a first large model training module, which is used to generate first sample query information based on sample knowledge data, and process the first sample query information through the initial large model to obtain training entity data; generate second sample query information based on the training entity data and sample requirement information, and process the second sample query information through the initial large model to obtain training key entity data; generate third sample query information based on the training key entity data, sample requirement information and sample entity constraint information, and process the third sample query information through the initial large model to obtain training table structures at different levels and hierarchical relationships between training table structures; construct a training data warehouse structure based on training table structures at different levels and hierarchical relationships between training table structures; adjust the parameters of the initial large model based on the training data warehouse structure and the label data warehouse structure until the initial large model converges to obtain a large model.

[0202] In some embodiments, the data warehouse structure generation device also includes: a data warehouse structure verification module, which is used to generate a fourth query information for data warehouse verification based on the data warehouse structure to be processed; and the fourth query information is responded to and processed through the large model to obtain the abnormal table structure in the data warehouse structure to be processed.

[0203] In some embodiments, the data warehouse structure generation device also includes: a second large model training module, which is used to modify the label data warehouse structure to obtain the abnormal data warehouse structure; generate a fourth sample query information based on the abnormal data warehouse structure; reply to the fourth sample query information through the large model to obtain a training abnormality table structure; determine the label abnormality table structure based on the difference between the label data warehouse structure and the abnormal data warehouse structure; adjust the parameters of the large model according to the training abnormality table structure and the label abnormality table structure until the parameters of the large model converge to obtain an updated large model.

[0204] Each module in the aforementioned data warehouse structure generation device may be implemented in whole or in part through software, hardware, or a combination thereof. Each module may be embedded in or independent of a processor in a computer device in the form of hardware, or may be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0205] In some embodiments, a computer device is provided. The computer device may be a server or a terminal. The computer device is described as a terminal. The internal structure diagram thereof may be as follows: Figure 15As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via wired or wireless communication, and the wireless communication can be achieved via WIFI, a mobile cellular network, NFC (near field communication), or other technologies. When executed by the processor, the computer program implements a method for generating a data warehouse structure. The display unit of the computer device is used to form a visually visible image, and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse, etc.

[0206] Those skilled in the art will understand that Figure 15 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0207] In some embodiments, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0208] Generate a first query message based on the business knowledge data, and query the business entity data based on the first query message; generate a second query message based on the business entity data and business demand information, and reply to the second query message to obtain the key entity data corresponding to the business demand information; generate a third query message based on the key entity data, business demand information and entity constraint information, and perform question-and-answer tasks based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures; construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between table structures.

[0209] In some embodiments, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0210] Generate a first query message based on the business knowledge data, and query the business entity data based on the first query message; generate a second query message based on the business entity data and business demand information, and reply to the second query message to obtain the key entity data corresponding to the business demand information; generate a third query message based on the key entity data, business demand information and entity constraint information, and perform question-and-answer tasks based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures; construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between table structures.

[0211] In some embodiments, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0212] Generate a first query message based on the business knowledge data, and query the business entity data based on the first query message; generate a second query message based on the business entity data and business demand information, and reply to the second query message to obtain the key entity data corresponding to the business demand information; generate a third query message based on the key entity data, business demand information and entity constraint information, and perform question-and-answer tasks based on the third query information to obtain table structures at different levels and hierarchical relationships between table structures; construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between table structures.

[0213] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0214] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, and the like.

[0215] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0216] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for generating a data warehouse structure, characterized in that: The method comprises: Generate first query information according to the knowledge data of the business, and query the entity data of the business based on the first query information; generating second query information based on the entity data and business requirement information of the business, and performing reply processing on the second query information to obtain key entity data corresponding to the business requirement information; Generating third query information based on the key entity data, the business requirement information, and the entity constraint information, and performing a question-answering task based on the third query information to obtain table structures at different levels and hierarchical relationships between the table structures; A data warehouse structure corresponding to the business demand information is constructed based on the table structures at different levels and the hierarchical relationships between the table structures.

2. The method according to claim 1, characterized in that The reply processing of the second query information to obtain key entity data corresponding to the business requirement information includes: Performing functional decomposition on the business requirement information included in the second query information to obtain sub-requirement information; From the entity data included in the second query information, key entity data corresponding to each sub-demand information is selected.

3. The method according to claim 1, characterized in that The performing of the question-answering task based on the third query information to obtain table structures at different levels and hierarchical relationships between the table structures includes: Selecting key entity data at different levels from the key entity data included in the third query information according to the business requirement information, and determining dependency relationships between the key entity data at different levels; Based on the key entity data at different levels, the dependency relationships, and the entity constraint information included in the third query information, table structures at different levels and hierarchical relationships between the table structures are determined.

4. The method according to claim 3, characterized in that The step of selecting key entity data at different levels from the key entity data included in the third query information according to the business requirement information and determining dependency relationships between the key entity data at different levels includes: Determine logic requirement information and display requirement information based on the business requirement information; selecting, from the key entity data included in the third query information, first key entity data belonging to a source level and second key entity data belonging to an aggregation level according to the logical requirement information, and determining a first dependency relationship between the first key entity data and the second key entity data; Among the key entity data, third key entity data belonging to the application layer is selected according to the display requirement information, and a second dependency relationship among the first key entity data, the second key entity data, and the third key entity data is determined.

5. The method according to claim 4, characterized in that The table structure includes a first table structure, a second table structure and a third table structure; the hierarchical relationship includes a first hierarchical relationship and a second hierarchical relationship; The determining of the table structures at different levels and the hierarchical relationships between the table structures based on the key entity data at different levels, the dependency relationships, and the entity constraint information included in the third query information includes: Constructing the first table structure belonging to the source level and the second table structure belonging to the aggregation level based on the first key entity data, the second key entity data, the first dependency relationship, and the entity constraint information included in the third query information; Determining a first hierarchical relationship between the first table structure and the second table structure based on the first dependency relationship; constructing the third table structure belonging to the application layer based on the first key entity data, the second key entity data, the third key entity data, the second dependency relationship and the entity constraint information; According to the second dependency relationship, a second hierarchical relationship among the first table structure, the second table structure, and the third table structure is determined.

6. The method according to claim 1, characterized in that The table structure includes a first table structure, a second table structure and a third table structure; The performing of the question-answering task based on the third query information to obtain table structures at different levels and hierarchical relationships between the table structures includes: Determine logic requirement information and display requirement information based on the business requirement information; Selecting first key entity data of a source level from the key entity data included in the third query information according to the logical requirement information; Constructing the first table structure belonging to the source hierarchy according to the first key entity data and the entity constraint information included in the third query information; Based on the first table structure, the logical requirement information and the display requirement information, the second table structure belonging to the aggregation level and the third table structure belonging to the application level are constructed, and the hierarchical relationship among the first table structure, the second table structure and the third table structure is determined.

7. The method according to claim 6, characterized in that The hierarchical relationship includes a first-level relationship and a second-level relationship; The step of constructing the second table structure belonging to the aggregation level and the third table structure belonging to the application level based on the first table structure, the logic requirement information, and the display requirement information, and determining the hierarchical relationship among the first table structure, the second table structure, and the third table structure includes: selecting a first candidate table structure from the first table structure according to the logical requirement information; Performing aggregation processing on the first candidate table structure to obtain the second table structure belonging to the aggregation level; selecting a second candidate table structure from the first table structure and the second table structure according to the display requirement information; Performing data extraction processing on the second candidate table structure to obtain the third table structure belonging to the application layer; The first hierarchical relationship between the first candidate table structure and the second table structure is determined, and the second hierarchical relationship between the second candidate table structure and the third table structure is determined.

8. The method according to claim 1, characterized in that The querying of the entity data of the business based on the first query information includes: Replying to the first query information through the big model to obtain entity data of the business; The reply processing of the second query information to obtain key entity data corresponding to the business requirement information includes: Replying to the second query information through the large model to obtain key entity data corresponding to the business requirement information; The performing of the question-answering task based on the third query information to obtain table structures at different levels and hierarchical relationships between the table structures includes: The question-answering task corresponding to the third query information is executed by the large model to obtain table structures at different levels and hierarchical relationships between the table structures.

9. The method according to claim 8, characterized in that The method further comprises: Generate first sample query information based on the sample knowledge data, and process the first sample query information using the initial large model to obtain training entity data; generating second sample query information based on the training entity data and sample requirement information, and processing the second sample query information through the initial large model to obtain training key entity data; generating third sample query information based on the training key entity data, the sample requirement information, and the sample entity constraint information, and processing the third sample query information using the initial large model to obtain training table structures at different levels and hierarchical relationships between the training table structures; Constructing a training data warehouse structure based on training table structures at different levels and hierarchical relationships between the training table structures; The parameters of the initial large model are adjusted according to the training data warehouse structure and the label data warehouse structure until the initial large model converges to obtain a large model.

10. The method according to claim 1, characterized in that The method further comprises: generating fourth query information for performing data warehouse verification according to the data warehouse structure to be processed; The fourth query information is responded to and processed by the large model to obtain the abnormal table structure in the data warehouse structure to be processed.

11. The method according to claim 10, characterized in that The method further comprises: Modify the label data warehouse structure to obtain the abnormal data warehouse structure; generating fourth sample query information based on the abnormal data warehouse structure; Replying to the fourth sample query information through the large model to obtain a training anomaly table structure; Determining a label anomaly table structure based on a difference between the label data warehouse structure and the anomaly data warehouse structure; The parameters of the large model are adjusted according to the training anomaly table structure and the label anomaly table structure until the parameters of the large model converge to obtain an updated large model.

12. A device for generating a data warehouse structure, characterized in that: The device comprises: an entity data determination module, configured to generate first query information according to the knowledge data of the business, and query the entity data of the business based on the first query information; a key entity data determination module, configured to generate second query information based on the entity data of the business and the business requirement information, and perform reply processing on the second query information to obtain key entity data corresponding to the business requirement information; a table structure and hierarchical relationship determination module, configured to generate third query information based on the key entity data, the business requirement information, and the entity constraint information, and perform a question-answering task based on the third query information to obtain table structures at different levels and hierarchical relationships between the table structures; The data warehouse structure construction module is used to construct a data warehouse structure corresponding to the business demand information based on the table structures at different levels and the hierarchical relationships between the table structures.

13. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 11 are implemented.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 11 are implemented.

15. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 11 are implemented.

Images