Enterprise data structured storage and query method, device and equipment applied to AI auxiliary total system

Abstract

The invention relates to the technical field of artificial intelligence, and discloses an enterprise data structured storage and query method, device and equipment applied to an AI auxiliary general system.The method comprises the steps that firstly, an enterprise data query task for the operation behavior of a target enterprise is determined; secondly, performing department matching on the first level of the enterprise knowledge base based on the enterprise data query task to obtain a target department object for executing the operation behavior; then, performing data matching on a second level of the enterprise knowledge base based on the enterprise data query task to obtain a target data object generated by the operation behavior; wherein the search range corresponding to the second hierarchy is smaller than the search range corresponding to the first hierarchy. And finally, outputting a query result corresponding to the enterprise data query task based on the target department object and the target data object. According to the method, cross-department enterprise data query can be realized, and the query accuracy and efficiency can be improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, and device for storing and querying structured enterprise data applied to an AI ​​deputy system. Background Art

[0002] Enterprise data management is a crucial component of modern business operations. In modern enterprises, it helps them efficiently manage vast amounts of data resources, support decision-making processes, and improve operational efficiency. With the proliferation of enterprise projects and documents, data complexity and diversity are also increasing, posing greater data management challenges for enterprises. Furthermore, the capabilities of relevant technologies to manage enterprise data still need to be further enhanced. Summary of the Invention

[0003] The present application provides a method, apparatus, and device for structured storage and query of enterprise data applied to an AI vice president system, which solves the technical problem in related technologies that the management capabilities of enterprise data still need to be improved, and achieves the technical effect of efficiently managing enterprise data.

[0004] In order to achieve the above objectives, the main technical solutions adopted in this application include:

[0005] In a first aspect, an embodiment of the present application provides a method for storing and querying enterprise data structured in an AI VP system. The enterprise data is structured and stored in an enterprise knowledge base. The method includes:

[0006] Determine the enterprise data query task targeting the target enterprise's operational behavior;

[0007] Based on the enterprise data query task, department matching is performed on the first level of the enterprise knowledge base to obtain a target department object for executing the operation behavior; wherein the enterprise knowledge base adopts a hierarchical structure to store enterprise data, and the hierarchical structure includes the first level and the second level, the first level is used to describe the organizational structure of the target enterprise, and the second level is used to describe the actual situation related to the operation behavior;

[0008] Performing data matching on the second level of the enterprise knowledge base based on the enterprise data query task to obtain a target data object generated by the operation behavior; wherein the search scope corresponding to the second level is smaller than the search scope corresponding to the first level;

[0009] The query result corresponding to the enterprise data query task is output based on the target department object and the target data object.

[0010] In a second aspect, an embodiment of the present application provides a device for storing and querying structured enterprise data applied to an AI VP system, wherein the enterprise data is structured and stored in an enterprise knowledge base, and the device includes:

[0011] A query task determination module is used to determine enterprise data query tasks targeting the operational behavior of a target enterprise;

[0012] a department matching module, configured to perform department matching on the first level of the enterprise knowledge base based on the enterprise data query task, and obtain a target department object for executing the operation behavior; wherein the enterprise knowledge base stores enterprise data in a hierarchical structure, the hierarchical structure comprising a first level and a second level, the first level being used to describe the organizational structure of the target enterprise, and the second level being used to describe actual conditions related to the operation behavior;

[0013] a data matching module, configured to perform data matching on the second level of the enterprise knowledge base based on the enterprise data query task to obtain a target data object generated by the operation behavior; wherein the search scope corresponding to the second level is smaller than the search scope corresponding to the first level;

[0014] A result output module is used to output the query result corresponding to the enterprise data query task based on the target department object and the target data object.

[0015] In a third aspect, an embodiment of the present application provides a computer device, including:

[0016] A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method described in any of the above embodiments by executing the computer instructions.

[0017] In an embodiment of the present application, first, based on the enterprise data query task, the target department object that performs the operational behavior is matched at the first level of the enterprise knowledge base, which can not only realize cross-departmental data query, but also quickly locate the department related to the operational behavior, thereby narrowing the query scope and improving query efficiency. Then, at the second level of the enterprise knowledge base, the target data object generated by the operational behavior is matched, and the specific data related to the operational behavior is further accurately locked to enhance the accuracy of the query. Finally, by integrating the information of the target department object and the target data object, a query result that is highly relevant to the enterprise data query task is output, ensuring the accuracy and relevance of the result, and providing reliable data support for the enterprise's operational decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the specific implementation methods of the present application or the technical solutions in the prior art, the following is a brief introduction to the drawings required for use in the specific implementation methods or the description of the prior art. Obviously, the drawings described below are some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0019] Figure 1 A flowchart of the enterprise data query method provided in the embodiments of this specification;

[0020] Figure 2 A flowchart of the enterprise data query method provided in the embodiments of this specification;

[0021] Figure 3 A flowchart of the enterprise data query method provided in the embodiments of this specification;

[0022] Figure 4 A flowchart of the enterprise data query method provided in the embodiments of this specification;

[0023] Figure 5 A flowchart of the enterprise data query method provided in the embodiments of this specification;

[0024] Figure 6 A flowchart of the enterprise data query method provided in the embodiments of this specification;

[0025] Figure 7 A flowchart of the enterprise data query method provided in the embodiments of this specification;

[0026] Figure 8 A schematic diagram of an enterprise data query device provided in an embodiment of this specification;

[0027] Figure 9 A schematic diagram of the structure of a computer device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0028] To make the purpose, technical solutions, and advantages of the embodiments of the present application more clear, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of this application.

[0029] With the rapid development of artificial intelligence (AI) technology, intelligent systems have become widely used to manage enterprise data. However, these systems suffer from the following drawbacks: insufficient processing capabilities for multi-source, heterogeneous data, unable to effectively handle documents from different departments and in different formats (such as PDF, Word, and Excel), leading to information silos and inefficient data utilization; limited knowledge management capabilities, often relying on keyword matching and traditional search methods, and unable to understand the deeper semantics of text; and low levels of intelligent task processing, lacking the automated capabilities to aggregate and analyze cross-departmental data and generate decision-support reports, impacting work efficiency.

[0030] Therefore, the present application provides a scenario example of an AI vice president system. Among them, the AI vice president system can be a system based on artificial intelligence, which can intelligently manage enterprise data like the vice president of the enterprise, provide query, analysis and decision support for enterprise data, thereby helping the general manager to manage the enterprise efficiently. First, the AI vice president system builds an enterprise knowledge base based on documents of different departments and different formats of the enterprise to realize hierarchical structured storage of enterprise data; then, the AI vice president system performs hierarchical intelligent query based on semantics in the enterprise knowledge base according to the enterprise data query task input by the user; finally, the AI vice president system performs intelligent analysis, generates reports, mind maps, etc. based on the query results.

[0031] This application also provides a scenario example of a method for building an enterprise knowledge base. First, the original enterprise documents of the target enterprise are collected. These original enterprise documents may include different formats (such as PDF, Word, Excel, PPT), different document structure types (such as linear structure, hierarchical structure, mesh structure, tree structure, modular structure, etc.) and different modal contents (such as text, table, image). Then, the original enterprise documents are pre-processed to convert them into text documents. For example, the initial text, table, image and other elements in the initial enterprise file can be identified and extracted using a parsing tool, and then the table is converted into JSON format text using a table to JSON tool. The image is converted into a text description using a text image multimodal large model, and the initial text and the converted text are saved in word format to form a standard enterprise file. Then, the outline planning agent is used to identify the key information in the standard enterprise file (such as author, file name, title and content keywords), analyze which department each file belongs to, and more levels of division can be performed under each department, and the enterprise file outline is constructed based on the analysis results. Next, the structure planning agent identifies the document structure type of each standard enterprise document. A pre-defined file structure type library can be created, storing the document structure types and characteristics of the target enterprise's standard enterprise documents. The structure planning agent compares the logical structure of the standard enterprise document with the structural characteristics in the file structure type library to determine the document structure type of each standard enterprise document. Next, the content planning agent parses the standard enterprise document according to its document structure type, converting it into hierarchically structured file data. Finally, the enterprise document outline and file data are hierarchically linked via file names to construct an enterprise knowledge base with hierarchical, structured storage.

[0032] In the above scenario example, multi-source heterogeneous data is converted into a unified standard enterprise file, and an enterprise knowledge base is built based on the standard enterprise file. This effectively solves the problem of insufficient multi-source heterogeneous data processing capabilities in related technologies, avoids information silos, and improves data utilization efficiency.

[0033] This application also provides a scenario example of an enterprise knowledge base query method. The enterprise knowledge base of the target enterprise is a hierarchical, structured storage knowledge base. The first level is the department level of the target enterprise, the second level is the document level of each department, the third level is the chapter level of each document, and the fourth level is the detailed content level of each chapter. During the query process, the user inputs an enterprise data query task related to the operational behavior of the target enterprise. The large model performs semantic understanding on the enterprise data query task, and then, based on the semantics of the enterprise data query task, first matches the target department at the first level, then matches the target document at the second level, then matches the target chapter at the third level, then matches the target content at the fourth level, and finally outputs the query results based on the semantic understanding of the target content. It should be noted that there may be more than one target queried at each level, so it is necessary to summarize each level before outputting the query results.

[0034] In the above scenario examples, based on the semantic understanding of user query tasks by the big model, the query process can accurately match the target content at each level. Compared with the use of keyword matching and simple retrieval in related technologies, it can better grasp the deep meaning of user queries and improve the accuracy of knowledge retrieval; through multi-level matching and aggregation, complex enterprise data query tasks are completed automatically, and information can be obtained and integrated across departments and documents to generate decision support results, which significantly improves work efficiency and intelligence level, and effectively solves the problem of low automation level of related technologies when handling complex tasks.

[0035] According to an embodiment of the present application, an embodiment of a method for structured storage and query of enterprise data applied to an AI vice president system is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0036] In this embodiment, a method for storing and querying enterprise data structured in an AI vice president system is provided. The enterprise data structured is stored in the enterprise knowledge base. Figure 1 , the method comprising:

[0037] S110. Determine an enterprise data query task targeting the operational behavior of the target enterprise.

[0038] The target enterprise can be an enterprise that needs to manage its data. Data management can include establishing an enterprise knowledge base based on the data source, querying the data, etc. Operational behavior can be the behavior of various activities in the enterprise involving sales, finance, marketing, etc. The enterprise data query task can be

[0039] A user-entered task to query data from an enterprise knowledge base.

[0040] In some embodiments, enterprise data query tasks can be directly entered into the enterprise knowledge base, such as "What was the company's total sales last year?" In other embodiments, the enterprise data query task is part of a complex task input by the user. For example, the enterprise knowledge base can work in conjunction with the process planning agent, and a complex task can be entered first, such as "Generate a sales report and mind map for the company for the past three years." The process planning agent can then perform in-depth analysis of the complex tasks input by the user based on the language capabilities of the large model, understand the intention behind the task, and decompose a complex task into multiple subtasks through intention decomposition. Some subtasks can be further divided into multiple steps. For example, the process planning agent can use the DeepSeek V3 large model and enter the following prompt words:

[0041] You are an intelligent assistant whose task is to understand the user's request and accurately parse their intent. Based on the information provided by the user, identify their main needs and break them down into actionable subtasks. If the task is complex, try breaking it down into multiple subtasks and listing the specific steps required for each subtask.

[0042] First, identify the task type. Based on the user input, identify the primary task type. Task types include: generating reports, analyzing data, creating mind maps, generating charts, and querying information. If the user request includes multiple tasks, identify all task types and label them.

[0043] Next, break down the task. For complex user requests, break them down into a series of subtasks. For each subtask, list the specific steps required. For example, 'Generate a sales report' can be broken down into steps like retrieving data, analyzing data, and generating a report.

[0044] For example, the process planning agent breaks down the task of "generating a company's sales report and mind map for the past three years" into two subtasks: generating a sales report for the past three years and generating a mind map of the company's sales data for the past three years. For example, the subtask "generating a company's sales report for the past three years" can be broken down into three steps: retrieving the company's sales data for the past three years, analyzing the data, and generating a report. The step "retrieving the company's sales data for the past three years" is an enterprise data query task.

[0045] S120 , performing department matching on the first level of the enterprise knowledge base based on the enterprise data query task to obtain a target department object for executing the operation behavior.

[0046] The enterprise knowledge base stores enterprise data in a hierarchical structure. The hierarchical structure consists of a first level and a second level. The first level describes the organizational structure of the target enterprise, while the second level describes the actual operational details. Department matching determines which departments the data being queried is relevant to. The target department object can be the department to which the data relevant to the enterprise data query belongs, which can be one or multiple.

[0047] In some embodiments, the organizational structure of the target enterprise includes sales department, marketing department, finance department, human resources department, production / operations department, R&D department, customer service department, procurement department, IT department, and legal department. The enterprise knowledge base of the target enterprise stores enterprise data in a hierarchical manner, and the first level is the name of the above departments.

[0048] In some implementations, a large model is used to perform semantic understanding on enterprise data query tasks, then matches them against the first-level department names in the enterprise knowledge base to obtain the target department object corresponding to the enterprise data query task. For example, if the enterprise data query task is "What was the revenue of the marketing department last year?", the large model can use semantic understanding to match the first-level department to the finance department.

[0049] In some implementations, the operational behaviors of each department are different. For example, the finance department is responsible for the company's financial reporting, taxation, budgeting, etc., and the sales department is responsible for the company's sales reporting, market analysis, sales strategy formulation, etc. Therefore, the second level of the enterprise knowledge base is described using operational behaviors.

[0050] In some embodiments, the second level may also include sub-levels, where the first sub-level is the name of the operating behavior of each department. For example, the first sub-level of the finance department is financial reporting, taxation, and budgeting. The second sub-level is the name of the standard enterprise documents contained in the first sub-level. For example, the second sub-level of the financial report is the first quarter financial statement of 2024 and the annual financial summary report of 2024. Furthermore, the content of the standard enterprise documents can be decomposed into multiple levels, such as three levels. The above three levels can correspond to the third sub-level, fourth sub-level, and fifth sub-level in the second level of the enterprise knowledge base.

[0051] S130 , performing data matching on the second level of the enterprise knowledge base based on the enterprise data query task to obtain target data objects generated by the operation behavior.

[0052] The search scope for the second level is smaller than that for the first level. Data matching can be performed by determining which data in the second level are related to the data being queried, such as data from a standard corporate document or a specific paragraph. The target data object can be one or more data objects that are required for the corporate data query task.

[0053] In some implementations, the second level of the enterprise knowledge base is described using operational behaviors. A large model is used to semantically understand enterprise data query tasks. After matching the target department object at the first level, data matching is performed at the second level corresponding to the target department object to obtain the target data object generated by the operational behaviors in the enterprise data query task. For example, if the enterprise data query task is "What was the revenue of R&D Department 2 last year?", data matching at the second level will yield the financial report.

[0054] In some embodiments, the second level of the enterprise knowledge base may also include sub-levels. For example, the first sub-level corresponding to the finance department is financial reports, taxes, and budget making. The second sub-level corresponding to the financial reports is the first quarter financial statements of 2024 and the 2024 annual financial summary report. The third sub-level corresponding to the 2024 annual financial summary report is the 2024 company financial summary, the financial situation of each department, summary and outlook. The fourth sub-level corresponding to the financial situation of each department is the financial situation of R&D Department 1, the financial situation of R&D Department 2..., and the large model is used to perform semantic understanding of the enterprise data query task. After matching the target department object at the first level, it is matched layer by layer in the sub-layers of the second level, and matched in turn to the financial report, the 2024 annual financial summary report, the financial situation of each department, and the financial situation of R&D Department 2. The financial situation of R&D Department 2 that is finally matched is the target data object.

[0055] S140: Output the query result corresponding to the enterprise data query task based on the target department object and the target data object.

[0056] In some embodiments, there is only one target department object and target data object. After obtaining the target department object and the target data object, a large model is used to perform semantic analysis on the target data object, and then semantic extraction is performed on the information most relevant to the enterprise data query task, and the information is output as the query result corresponding to the enterprise data query task.

[0057] In some implementations, there is one target department object and multiple target data objects. A large model is first used to perform semantic analysis on each target data object. Then, the most relevant information for the enterprise data query task is semantically extracted to obtain a query result for the target data object. These query results are then aggregated and output as the query result for the target department object, which also serves as the query result for the enterprise data query task.

[0058] In some embodiments, there are multiple target department objects, each corresponding to a plurality of target data objects. First, a semantic analysis is performed on each target data object using a large model. Then, semantic extraction is performed on the information most relevant to the enterprise data query task to obtain a query result for the target data object. The query results for the target data objects corresponding to each target department object are then aggregated to form the query result for the corresponding target department object. Finally, the query results for each target department object are aggregated and output as the query result for the enterprise data query task.

[0059] In the above embodiment, first, based on the enterprise data query task, the target department object that performs the operational behavior is matched at the first level of the enterprise knowledge base. This not only enables cross-departmental data queries, but also allows for the rapid location of departments related to the operational behavior, thereby narrowing the query scope and improving query efficiency. Next, at the second level of the enterprise knowledge base, the target data object generated by the operational behavior is matched, further accurately targeting specific data related to the operational behavior and enhancing the accuracy of the query. Finally, by integrating the information of the target department object and the target data object, a query result highly relevant to the enterprise data query task is output, ensuring the accuracy and relevance of the result, and providing reliable data support for the enterprise's operational decision-making.

[0060] In some embodiments, the second level includes a file layer; data matching is performed on the second level of the enterprise knowledge base based on the enterprise data query task to obtain the target data object generated by the operational behavior, including: file matching is performed on the file layer of the enterprise knowledge base based on the enterprise data query task to obtain the target file object generated by the operational behavior.

[0061] The target data object includes a target file object. File matching can be used to determine which standard enterprise file the data to be queried is related to. The target file object can be one or more files related to the data to be queried in the enterprise data query task.

[0062] In some implementations, the first level of the enterprise knowledge base consists of department names, and the second level consists of a file layer described using the names of standard enterprise files. A large model is used to semantically understand enterprise data query tasks. After matching the target department object at the first level, file matching continues at the second level corresponding to the target department object to obtain the target file object generated by the operational behavior in the enterprise data query task. For example, the enterprise data query task is "What was the revenue of R&D Department 2 last year?", and the second level matches the 2024 annual financial summary report.

[0063] In some embodiments, the first level of the enterprise knowledge base is the department name, and the second level may also include sub-levels. For example, the first sub-level is a file level described using the name of a standard enterprise file. Furthermore, the content of the standard enterprise file may be decomposed into multiple levels, such as two levels. The two levels may correspond to the second sub-level and the third sub-level in the second level of the enterprise knowledge base. A large model is used to perform semantic understanding of enterprise data query tasks. After matching the target department object at the first level, file matching is continued at the first sub-level of the second level corresponding to the target department object to obtain the target file object generated by the operational behavior in the enterprise data query task. Furthermore, data matching may be continued at the second sub-level and the third sub-level to obtain a more accurate target data object.

[0064] See also Figure 2 In some embodiments, the hierarchical structure further includes a third level and a fourth level, and the search ranges corresponding to the first level, the second level, the third level, and the fourth level decrease in sequence; the third level is used to describe the file structure of the target file object, and the fourth level is used to describe the specific content of the target file object; before outputting the query results corresponding to the enterprise data query task based on the target department object and the target data object, the method further includes:

[0065] S310 , performing file structure matching on the third level of the enterprise knowledge base based on the enterprise data query task to obtain a target file structure of the target file object.

[0066] S320: Perform content matching on the fourth level of the enterprise knowledge base based on the enterprise data query task to obtain target content of the target file object.

[0067] The file structure can be a standard enterprise file structure with sections, paragraphs, and other hierarchies. File structure matching can determine which parts of the standard enterprise file structure are relevant to the data being queried. The target file structure can be one or more file structures related to the data being queried in the enterprise data query task.

[0068] Content matching can be used to determine which file contents are related to the data being queried. The target content can be one or more file contents related to the data being queried for the enterprise data query task.

[0069] In some implementations, the first level of the enterprise knowledge base is department names, the second level is a file layer described using standard enterprise file names, the third level is a file structure layer, and the fourth level is a specific content layer. Using a large model to perform semantic understanding of enterprise data query tasks, the first level matches the target department object; then, a file match is performed on the second level corresponding to the target department object to obtain the target file object; then, a file structure match is performed on the file structure layer corresponding to the target file object to obtain the target file structure; and finally, a content match is performed on the specific content layer corresponding to the target file structure to obtain the target content. For example, the enterprise data query task is "What was the revenue of the marketing department last year?". The big model is used to perform semantic understanding of the enterprise data query task. First, it matches the finance department at the first level; then performs file matching at the second level corresponding to the finance department to obtain the "2024 Annual Financial Report"; then performs file structure matching at the file structure layer corresponding to the 2024 Annual Financial Report to obtain "Chapter 2: Department Financial Situation"; finally, performs content matching at the specific content layer corresponding to Chapter 2 to obtain "Marketing Department Revenue: The marketing department's revenue in 2024 reached 120 million yuan, a year-on-year increase of 10%."

[0070] S330. Output the query result corresponding to the enterprise data query task based on the target department object, target file object, target file structure and target content.

[0071] In some embodiments, one target file object corresponds to multiple target file structures, and one target file structure corresponds to multiple target contents. First, a large model is used to perform semantic analysis on each target content, and then semantic extraction is performed on the information most relevant to the enterprise data query task to obtain the query result of the target content; then the query results corresponding to the target content are aggregated to the corresponding target file structure as the query result of the target file structure; finally, the query results of each target file structure are aggregated to the corresponding target file object as the query result of the target file object.

[0072] In some embodiments, there are multiple target department objects, each corresponding to a plurality of target data objects. First, a semantic analysis is performed on each target data object using a large model. Then, semantic extraction is performed on the information most relevant to the enterprise data query task to obtain a query result for the target data object. The query results for the target data objects corresponding to each target department object are then aggregated to form the query result for the target department object. Finally, the query results for each target department object are aggregated and output as the query result corresponding to the enterprise data query task.

[0073] See also Figure 3 In some embodiments, the enterprise knowledge base is constructed by:

[0074] S410: Construct an outline planning agent, a structure planning agent, and a content planning agent.

[0075] In some implementations, building an enterprise knowledge base can be accomplished through a series of intelligent steps. Specifically, it is necessary to construct an outline planning agent, a structure planning agent, and a content planning agent. These agents can utilize prompts and the capabilities of the large model to complete their respective tasks.

[0076] S420. Construct an enterprise document outline for the target enterprise through an outline planning intelligent agent.

[0077] The enterprise document outline is constructed based on standard enterprise documents in a specified format. Specifically, standard enterprise documents can be documents that meet the established specifications of the target enterprise. These documents have unified standards in format and content organization, making them easier for intelligent agents to parse and process.

[0078] In some embodiments, the standard enterprise document is a plain text file obtained by processing the initial enterprise document. The initial enterprise document contains text, tables, and images. A table-to-JSON tool can be used to convert the table into JSON format. The image can be converted into a text description using the Text-Image Multimodal Macro Model, and image tags can be added to the converted text description. Simultaneously, the original image is saved to a pre-defined storage space, and an image link is inserted into the image's tag area.

[0079] In some implementations, the outline planning agent analyzes key information in standard enterprise documents, categorizes them according to pre-set classification rules, and constructs a hierarchical enterprise document outline. For example, the first level of the enterprise document outline consists of the target enterprise's department names, the second level consists of each department's operational activities (e.g., the finance department's operational activities include financial reporting, taxation, and budgeting), and the third level consists of the standard enterprise document file names.

[0080] S430. Perform structural analysis on the standard enterprise file through the structure planning intelligent agent to obtain a target structure type of the standard enterprise file.

[0081] Among them, structural analysis can be to analyze the text content of standard enterprise documents and identify the structural features of the documents.

[0082] In some embodiments, the structural planning agent can match the structural analysis results of the standard enterprise file with the characteristics of multiple preset file structure types (such as hierarchical structure, network structure, linear structure, hybrid structure, etc.) to obtain the target structure type of the standard enterprise file.

[0083] S440. Decompose the standard enterprise files into layers through the content planning agent to obtain file granularity level data of the standard enterprise files.

[0084] In some embodiments, the content planning agent can hierarchically decompose the standard enterprise file into multiple levels based on the target structure type of the standard enterprise file (for example, decomposing the hierarchical file into three levels: chapter, section, and paragraph); and then perform semantic extraction on different levels to obtain file granularity level data of the standard enterprise file.

[0085] S450: Construct a knowledge base based on the enterprise document outline and document granularity level data to obtain an enterprise knowledge base.

[0086] In some embodiments, the enterprise document outline includes multiple levels, where the first level is the department name of the target enterprise, and the last level is the file name of the standard enterprise document, which is marked as the Mth level. The file granularity level data includes N levels. It is understood that the enterprise document outline and the file granularity level data can be hierarchically connected through the file name to construct an enterprise knowledge base with M+N levels.

[0087] In the above embodiment, first, an outline planning agent, a structure planning agent and a content planning agent are constructed to provide core tools for the construction of an enterprise knowledge base; then, an enterprise document outline of the target enterprise is constructed through the outline planning agent to provide a high-level data foundation for the construction of an enterprise knowledge base; then, the standard enterprise document is structurally analyzed through the structure planning agent to obtain the target structure type of the standard enterprise document, and the standard enterprise document is hierarchically decomposed through the content planning agent to obtain the file granularity level data of the standard enterprise document, providing a low-level data foundation for the construction of an enterprise knowledge base; finally, a knowledge base is constructed based on the enterprise document outline and the file granularity level data to obtain an enterprise knowledge base, which provides the target enterprise with a cross-departmental, hierarchical and accurate retrieval-supporting knowledge management platform.

[0088] See also Figure 4In some embodiments, constructing an enterprise document outline of a target enterprise by an outline planning agent includes:

[0089] S510: Determine the preset architecture data of the target enterprise.

[0090] The pre-set schema data is used to represent the attributable dimensions of standard enterprise documents. It can include a single level (e.g., the target enterprise's departmental divisions) or multiple levels (e.g., the first level is the departmental divisions, and the second level is the multiple operational activities of each department).

[0091] In some embodiments, the preset architecture data may be dimensional information pre-defined by the target enterprise and used to classify and attribute standard enterprise files. For example, the preset architecture data of a manufacturing enterprise includes dimensions such as "Product R&D Department", "Production Management Department", and "Quality Control Department" for classifying standard enterprise files.

[0092] S520: Obtain the attributes to be identified of each standard enterprise file.

[0093] Among them, the attributes to be identified include the author, file name, title and content keywords of the standard enterprise file.

[0094] In some implementations, a large model is used to capture information such as the author, file name, title, and content keywords of standard corporate documents. For example, a document might be authored by "Zhang San," with a file name of "2024 First Quarter Sales Report" and a title of "Sales Performance Analysis." Content keywords might include "sales," "customer list," and "market trends." These attributes reflect the content characteristics and creation context of standard corporate documents, providing key information for subsequent file attribution analysis.

[0095] S530: Determine the attributable object data of each standard enterprise file based on the preset architecture data and the attributes to be identified.

[0096] The attributable object data may be information obtained based on analysis of the preset architecture data and attributes to be identified, indicating a specific object in the preset architecture data to which the standard enterprise file may be attributed, such as the sales department.

[0097] In some implementations, the to-be-identified attributes of a document are matched against pre-set schema data to determine the most appropriate attribution dimension for the document. For example, if the author of a document belongs to the "Sales Department," the document name and title both contain "sales"-related words, and the content keywords are closely related to sales, then by matching the departmental division dimension in the pre-set schema data, the attributable object data for this document can be determined to be the "Sales Department."

[0098] S540: Construct an enterprise document outline for the target enterprise based on the document identification data and attributable object data of each standard enterprise document.

[0099] The file identification data may be information used to uniquely identify each standard enterprise file, such as the file's unique number, name, creation time, etc.

[0100] In some embodiments, the outline planning agent is driven by a large language model (such as DeepSeek V3) and constructs the enterprise document outline based on the designed system prompt words and task prompt words. For example, the system prompt words are as follows:

[0101] Role: You are the "Outline Planner" in the AI VP system, focused on analyzing corporate documents and building document outlines. Based on the document's content, keywords, context, author, and other information, you need to determine the department or business area to which it belongs, and organize it into a clear hierarchical structure.

[0102] Objective: Your goal is to help management quickly locate and retrieve documents across the company, improving knowledge management efficiency. You will ensure accurate categorization, a clear hierarchy, and compliance with the company's organizational structure.

[0103] The task prompts are as follows:

[0104] ① Read and understand the document content, including the text, title, author information, time and other metadata.

[0105] ② Extract keywords, analyze core topics, and determine which department or business area the document belongs to.

[0106] ③ Based on the organizational logic of corporate documents, build a hierarchical outline to make its structure clear and expandable.

[0107] ④ Output clear file classification results and outline structure in JSON format.

[0108] Example of a constructed enterprise document outline:

[0109]

[0110]

[0111] In the above embodiment, first, the preset architecture data of the target enterprise is determined, which provides a basic framework for the construction of the enterprise file outline, ensuring that standard enterprise files can be organized and managed according to the preset architecture; then, the attributes to be identified of each standard enterprise file are obtained, providing a basis for subsequent file attribution judgment; then, based on the preset architecture data and the attributes to be identified, the attributable object data of each standard enterprise file is determined, thereby ensuring that the file can be correctly classified to facilitate subsequent management; finally, based on the file identification data and attributable object data of each standard enterprise file, the enterprise file outline of the target enterprise is constructed, which can improve the efficiency and accuracy of enterprise file management and provide a solid foundation for the construction of the enterprise knowledge base.

[0112] See also Figure 5 In some embodiments, a structure planning agent performs structural analysis on a standard enterprise document to obtain a target structure type of the standard enterprise document, including:

[0113] S610. Build a file structure type library for the target enterprise.

[0114] The file structure type library stores preset file structure types and structural features of the preset file structure types.

[0115] In some implementations, by analyzing the target enterprise's file usage habits and business needs, the most commonly used file structure types in the target enterprise are determined, and their characteristics are recorded in detail in a file structure type library. For example, the file structure type library is as follows:

[0116] 1) Linear structure: The contents of the document are arranged in chronological or logical order, and are expanded from the beginning to the end. Common document types include reports (such as work summaries, project reports, etc.), notices (issued in chronological or event order), meeting minutes (recorded according to meeting procedures), instructions (described in steps or processes),

[0117] 2) Hierarchical Structure: Document content is organized into distinct hierarchies, typically consisting of titles, subtitles, and paragraphs. Common document types include regulations (e.g., company charters and employee handbooks), legal documents (e.g., contracts and agreements), technical documents (e.g., operating manuals and development documentation), and scientific papers (e.g., abstract, introduction, methods, results, and discussion).

[0118] 3) Network structure: File contents are interconnected through hyperlinks or references, forming a complex network. Common file types include web pages (such as HTML files, which use hyperlinks to jump to other files), knowledge bases (such as Wiki documents, which reference each other), and project management tools (such as Gantt charts and task dependency diagrams).

[0119] 4) Table Structure: File contents are organized into rows and columns for easy comparison and analysis. Common file types include financial statements (such as balance sheets and income statements), statistical tables (such as sales data and inventory tables), and schedules (such as meeting schedules and project progress sheets).

[0120] 5) Graphical Structure: Documents primarily use graphics, charts, or images to visually display information. Common document types include flowcharts (e.g., business processes and workflows), organizational charts (showing departmental or personnel relationships), statistical charts (e.g., bar charts, pie charts, line graphs), and design drawings (e.g., engineering drawings and product designs).

[0121] 6) Tree Structure: File content is organized in a tree-like, branching format, suitable for classification and hierarchical display. Common file types include directory files (such as file directories and data indexes), classification documents (such as product classifications and knowledge classifications), and decision trees (such as risk assessments and decision analysis).

[0122] 7) Modular Structure: File content is divided into multiple independent modules, each of which performs a specific function or describes a specific content. Common file types include software documentation (such as API documentation and user manuals), training materials (such as modular training courses), and project plans (such as modular task assignments).

[0123] 8) Hybrid structure: A document combines multiple structural forms and is flexibly organized according to the content. Common document types include comprehensive reports (such as annual reports, which contain text, tables, charts, etc.) and business plans (which combine linear narrative, tabular data, and graphical presentation).

[0124] 9) Database Structure: Files are stored in a database format, with content organized into fields and records. Common file types include customer information databases (such as customer data in a CRM system), inventory management systems (such as merchandise inventory records), and personnel files (such as employee information databases).

[0125] 10) Timeline Structure: This structure organizes content chronologically, making it suitable for presenting history or plans. Common file types include project schedules (which display task progress over time), historical records (such as a company's development history), and event logs (such as system logs and operation records).

[0126] 11) Question-and-Answer Structure: This document organizes content into questions and answers, ideal for solving specific problems. Common document types include FAQs (Frequently Asked Questions) and technical support documents (such as troubleshooting guides).

[0127] 12) Free-form structure: There is no fixed structure, and the content is freely organized according to needs. Common file types include creative copy (such as advertising copy, promotional materials) and casual records (such as meeting notes, inspiration records).

[0128] S620: Perform structural analysis on the standard enterprise file to obtain file structure features of the standard enterprise file.

[0129] Among them, structure analysis can be a process of analyzing standard enterprise files to extract their file structure features.

[0130] In some implementations, standard enterprise documents incorporate special markup into the text resulting from the conversion of tables and images from the original enterprise documents. By performing natural language processing on the content of these standard enterprise documents and analyzing the organization of elements such as titles, paragraphs, table tags, and image tags, the file structure characteristics of the standard enterprise documents are derived.

[0131] S630: Match the structural features of the preset file structure type with the file structure features, and determine the target structure type of each standard enterprise file in the preset file structure types in the file structure type library.

[0132] The matching can be a process of comparing the file structure features of the standard enterprise file with the structure features of the preset file structure type in the file structure type library. Specifically, the matching can be performed by calculating the similarity between the file structure features and the preset file structure type features.

[0133] In some embodiments, the structure planning agent is driven by a large language model (e.g., DeepSeek V3) and identifies the target structure type of a standard enterprise document based on pre-designed system prompts and task prompts. For example, the system prompts are as follows:

[0134] Role: You are the "Structural Planner" in the AI VP system, focusing on analyzing the organizational structure of corporate documents and identifying their structural types. Your goal is to accurately determine the structural category of documents based on the capabilities of the local structure library and large language models, making subsequent knowledge extraction and organization more efficient.

[0135] The task prompts are as follows:

[0136] ① Analyze the content format of the document, including titles, paragraphs, lists, tables, hyperlinks, images and other information.

[0137] ② Combined with the classification rules of the local structure library, through the semantic analysis of the large model, analyze the structural category to which the file belongs and output the final result.

[0138] In the above embodiment, a file structure type library of the target enterprise is first constructed to provide a comparison basis for subsequent standard enterprise file structure matching; then, a structural analysis is performed on the standard enterprise file to obtain the file structure characteristics of the standard enterprise file, providing data for subsequent standard enterprise file structure matching; finally, the structural characteristics of the preset file structure type are matched with the file structure characteristics to determine the target structure type of each standard enterprise file, thereby providing a file structure type basis for subsequent processing of the standard enterprise file.

[0139] See also Figure 6 In some embodiments, a content planning agent performs hierarchical decomposition of standard enterprise files to obtain file granularity level data of the target enterprise, including:

[0140] S710: Determine preset granularity data of the target enterprise.

[0141] The preset granularity data is used to characterize the granularity into which standard enterprise files can be divided.

[0142] In some embodiments, the target enterprise sets preset granularity data based on the level of detail and management requirements of the standard enterprise document content, including different levels of division granularity such as chapters, paragraphs, tables, and images, to facilitate more detailed management of the content of the standard enterprise document.

[0143] S720: Determine the file granularity level data of the standard enterprise file based on the preset granularity data and the target structure type.

[0144] File granularity data can be the semantic extraction results of standard enterprise files at different levels, reflecting the key information of the file content at each granularity level. The determination of file granularity data allows the file content to be broken down into multiple levels, facilitating subsequent storage and retrieval.

[0145] In some implementations, the content planning agent is driven by a large language model (e.g., DeepSeek V3) and constructs file-level granularity data based on pre-designed system prompts and task prompts. For example, the system prompts are as follows:

[0146] Role: You are the "Content Planner" in the AI VP system, focusing on parsing the content of corporate documents and performing semantic extraction. Your goal is to leverage the power of large language models to accurately parse corporate documents, efficiently extract key information, and convert it into a structured data format to better support corporate knowledge management and decision-making analysis.

[0147] The task prompts are as follows:

[0148] ① Obtain document structure information: call the recognition results of the structure planner to determine which file structure type the document belongs to.

[0149] ② Content decomposition: Decompose the text according to the document’s file structure type (chapter, title, table, etc.), and perform a fine-grained analysis of each part.

[0150] ③Semantic extraction: Use large models to extract core concepts, definitions, and relationships to form structured information.

[0151] ④ Unified mapping: All content is converted into a hierarchical knowledge tree to facilitate knowledge base retrieval.

[0152] For example, the content of the standard enterprise document is as follows:

[0153] 2024 Annual Company Operation Summary

[0154] Performance Overview: The company's revenue reached 500 million yuan this year, a 15% year-on-year increase. (Attached is a bar chart showing revenue growth by department)

[0155] Market Analysis: Market share increased to 25%, with North America showing the fastest growth. (Attached table: Market share)

[0156] Future Plan: New products will be launched in 2025 to expand into overseas markets. (Attached is a flowchart: Product Launch Timeline) The resulting file granularity data is as follows:

[0157]

[0158]

[0159] In the above embodiment, the preset granularity data of the target enterprise is first determined to provide a reference standard for the hierarchical disassembly of the standard enterprise file; then, based on the preset granularity data and the target structure type, the file granularity hierarchical data of the standard enterprise file is determined, thereby dividing the content of the standard enterprise file according to the established granularity level and file structure type, making the file granularity hierarchical data more standardized.

[0160] See also Figure 7 In some embodiments, the standard enterprise file is obtained by:

[0161] S810: Extract elements from the initial enterprise file to obtain initial text elements and initial non-text elements.

[0162] S820: Convert the initial non-text element into text to obtain a converted text element.

[0163] S830. Save the initial text element and the converted text element according to the specified format to obtain a standard enterprise file.

[0164] Initial corporate files can be original documents stored in various formats (such as PDF, Word, Excel, PPT, etc.) by different departments of the target enterprise. For example, financial reports or meeting minutes that have not undergone any format standardization may contain complex typesetting, embedded images, or non-text elements such as tables. Element extraction can be the process of identifying and separating text and non-text elements from initial corporate files. Initial text elements can be text content such as titles and paragraphs in the initial corporate files. Initial non-text elements can be non-text information such as tables, images, etc. in the initial corporate files.

[0165] Text conversion can be a process of converting an initial non-text element into a text form. The converted text element can be a text element obtained by converting a non-text element.

[0166] The specified format can be a unified format pre-defined by the target enterprise for constructing standard enterprise files, such as using word or txt format. In addition, the converted text elements are marked, and image links are added to the converted text elements obtained by image conversion, etc., so as to be used when tables and images need to be identified in subsequent file structure type identification.

[0167] In some implementations, a parsing tool (such as Apache Tika) is first used to identify and extract text, tables, images, and other elements from the original enterprise document. A table-to-JSON tool is then used to convert the tables into JSON format. The images are then converted into text descriptions using a large multimodal text-image model. Tags are then added to the converted JSON-formatted text and image descriptions. Simultaneously, the original images are saved to a pre-defined storage space, and links to them are inserted into the corresponding tagged areas. Finally, the original and converted text elements are saved in Word format to produce a standard enterprise document.

[0168] In the above embodiment, converting the initial enterprise file into a standard enterprise file ensures uniform formatting and accessibility of the content, providing a foundation for subsequent agents (such as the outline planning agent, structure planning agent, and content planning agent) to further process and analyze the file. This not only improves file processing efficiency but also enhances the practicality of the enterprise knowledge base.

[0169] See also Figure 8 In an embodiment of the present application, a device 900 for storing and querying structured enterprise data applied to an AI vice president system is provided. The device 900 for storing and querying structured enterprise data applied to an AI vice president system includes:

[0170] A query task determination module 910 is used to determine an enterprise data query task targeting the operation behavior of a target enterprise;

[0171] Department matching module 920 is used to perform department matching on the first level of the enterprise knowledge base based on the enterprise data query task to obtain the target department object for executing the operation behavior. The enterprise knowledge base adopts a hierarchical structure to store enterprise data. The hierarchical structure includes a first level and a second level. The first level is used to describe the organizational structure of the target enterprise, and the second level is used to describe the actual situation related to the operation behavior.

[0172] Data matching module 930 is used to perform data matching on the second level of the enterprise knowledge base based on the enterprise data query task to obtain target data objects generated by operational behaviors; wherein the search scope corresponding to the second level is smaller than the search scope corresponding to the first level;

[0173] The result output module 940 is used to output the query result corresponding to the enterprise data query task based on the target department object and the target data object.

[0174] In some implementations, the second level includes a file level; the data matching module 930 further includes:

[0175] The file matching unit is used to perform file matching on the file layer of the enterprise knowledge base based on the enterprise data query task to obtain the target file object generated by the operation behavior; wherein the target data object includes the target file object.

[0176] In some embodiments, the hierarchical structure further includes a third level and a fourth level, and the search ranges corresponding to the first level, the second level, the third level, and the fourth level decrease in sequence; the third level is used to describe the file structure of the target file object, and the fourth level is used to describe the specific content of the target file object; before the result output module 940, the device 900 for storing and querying enterprise data structured in the AI VP system further includes:

[0177] A structure matching module is used to perform file structure matching on the third level of the enterprise knowledge base based on the enterprise data query task to obtain the target file structure of the target file object;

[0178] A content matching module is used to perform content matching on the fourth level of the enterprise knowledge base based on the enterprise data query task to obtain the target content of the target file object;

[0179] The second result output module is used to output the query result corresponding to the enterprise data query task based on the target department object, the target file object, the target file structure and the target content.

[0180] In some embodiments, the device 900 for storing and querying structured enterprise data in an AI VP system further includes:

[0181] An agent building module, used to build an outline planning agent, a structure planning agent, and a content planning agent;

[0182] An outline construction module is used to construct an enterprise document outline of a target enterprise through an outline planning intelligent agent; wherein the enterprise document outline is constructed based on a standard enterprise document in a specified format;

[0183] The structure analysis module is used to analyze the structure of the standard enterprise file through the structure planning agent to obtain the target structure type of the standard enterprise file;

[0184] The file disassembly module is used to perform hierarchical disassembly of standard enterprise files through the content planning agent to obtain file granularity level data of the standard enterprise files;

[0185] The knowledge base construction module is used to construct the knowledge base based on the enterprise document outline and document granularity level data to obtain the enterprise knowledge base.

[0186] In some implementations, the outline building module further includes:

[0187] A preset architecture determining unit, configured to determine preset architecture data of a target enterprise; wherein the preset architecture data is used to represent attributable dimensions of a standard enterprise document;

[0188] A file attribute acquisition unit is used to acquire the attributes to be identified of each standard enterprise file; wherein the attributes to be identified include the author, file name, title and content keywords of the standard enterprise file;

[0189] an attributable object determining unit, configured to determine attributable object data of each standard enterprise file based on preset architecture data and attributes to be identified;

[0190] The outline construction unit is used to construct the enterprise file outline of the target enterprise based on the file identification data and the attributable object data of each standard enterprise file.

[0191] In some embodiments, the structure analysis module further comprises:

[0192] A type library construction unit is used to construct a file structure type library of a target enterprise; wherein the file structure type library stores preset file structure types and structural features of the preset file structure types;

[0193] A structure analysis unit is used to perform structure analysis on standard enterprise files to obtain file structure features of the standard enterprise files;

[0194] The feature matching unit is used to match the file structure features based on the structure features of the preset file structure type, and determine the target structure type of each standard enterprise file in the preset file structure types in the file structure type library.

[0195] In some implementations, the file decomposition module further includes:

[0196] A preset granularity determination unit is used to determine preset granularity data of a target enterprise; wherein the preset granularity data is used to characterize the granularity into which a standard enterprise file can be divided;

[0197] The granularity data determining unit is used to determine the file granularity level data of the standard enterprise file based on the preset granularity data and the target structure type.

[0198] In some embodiments, the device 900 for storing and querying structured enterprise data in an AI VP system further includes:

[0199] An element extraction module is used to extract elements from the initial enterprise document to obtain initial text elements and initial non-text elements;

[0200] A text conversion module, used for converting the initial non-text element into text to obtain a converted text element;

[0201] The file saving module is used to save the initial text elements and the converted text elements in a specified format to obtain a standard enterprise file.

[0202] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0203] The enterprise data query device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0204] See also Figure 9 , Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. Figure 9As shown, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in the memory or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 9 A processor 10 is taken as an example.

[0205] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0206] The memory 20 stores instructions that can be executed by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0207] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0208] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0209] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 20 can be connected via a bus or other means. Figure 9 The bus connection is taken as an example.

[0210] The input device 30 can receive input digital or character information and generate key signal input related to user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a trackpad, a touch pad, an indicator stick, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 can include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor). The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display, and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0211] The embodiments of the present application also provide a computer-readable storage medium. The above-mentioned method according to the embodiment of the present application can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0212] An embodiment of the present application provides a computer program product, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform a method according to any embodiment of the present application.

[0213] Although the embodiments of the present application have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations shall fall within the scope defined by the appended claims.

[0214] The systems, devices, modules, or units described in the above embodiments may be implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

[0215] For the convenience of description, the above devices are described as being divided into various units according to their functions. Of course, when implementing this application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0216] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0217] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0218] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0219] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0220] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0221] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0222] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

[0223] Although the embodiments of the present application have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present application, and such modifications and variations shall fall within the scope defined by the appended claims.

Claims

1. A method for storing and querying enterprise data structured in an AI VP system, characterized in that: The enterprise data is structured and stored in an enterprise knowledge base. The method includes: Determine the enterprise data query task targeting the target enterprise's operational behavior; Based on the enterprise data query task, department matching is performed on the first level of the enterprise knowledge base to obtain a target department object for executing the operation behavior; wherein the enterprise knowledge base adopts a hierarchical structure to store enterprise data, and the hierarchical structure includes the first level and the second level, the first level is used to describe the organizational structure of the target enterprise, and the second level is used to describe the actual situation related to the operation behavior; Performing data matching on the second level of the enterprise knowledge base based on the enterprise data query task to obtain a target data object generated by the operation behavior; wherein the search scope corresponding to the second level is smaller than the search scope corresponding to the first level; The query result corresponding to the enterprise data query task is output based on the target department object and the target data object.

2. The method according to claim 1, characterized in that The second level includes a file level; performing data matching on the second level of the enterprise knowledge base based on the enterprise data query task to obtain a target data object generated by the operation behavior includes: Based on the enterprise data query task, file matching is performed on the file layer of the enterprise knowledge base to obtain a target file object generated by the operation behavior; wherein, the target data object includes the target file object.

3. The method according to claim 2, characterized in that The hierarchical structure further includes a third level and a fourth level, wherein the search ranges corresponding to the first level, the second level, the third level, and the fourth level are successively reduced; the third level is used to describe the file structure of the target file object, and the fourth level is used to describe the specific content of the target file object; Before outputting the query result corresponding to the enterprise data query task based on the target department object and the target data object, the method further includes: Performing file structure matching on the third level of the enterprise knowledge base based on the enterprise data query task to obtain a target file structure of the target file object; Performing content matching on the fourth level of the enterprise knowledge base based on the enterprise data query task to obtain target content of the target file object; The outputting the query result corresponding to the enterprise data query task based on the target department object and the target data object includes: The query result corresponding to the enterprise data query task is output based on the target department object, the target file object, the target file structure and the target content.

4. The method according to any one of claims 1 to 3, characterized in that The enterprise knowledge base is constructed in the following ways: Construct outline planning agents, structure planning agents, and content planning agents; Constructing an enterprise document outline of the target enterprise through the outline planning agent; wherein the enterprise document outline is constructed based on a standard enterprise document in a specified format; Performing structural analysis on the standard enterprise file by the structure planning agent to obtain a target structure type of the standard enterprise file; The standard enterprise file is hierarchically disassembled by the content planning agent to obtain file granularity level data of the standard enterprise file; A knowledge base is constructed based on the enterprise document outline and the document granularity level data to obtain the enterprise knowledge base.

5. The method according to claim 4, characterized in that The step of constructing the enterprise document outline of the target enterprise by the outline planning agent includes: Determining preset architecture data of the target enterprise; wherein the preset architecture data is used to characterize the attributable dimensions of the standard enterprise document; Obtaining attributes to be identified of each standard enterprise file; wherein the attributes to be identified include the author, file name, title, and content keywords of the standard enterprise file; Determining attributable object data for each standard enterprise file based on the preset architecture data and the attributes to be identified; The enterprise document outline of the target enterprise is constructed based on the document identification data of each standard enterprise document and the attributable object data.

6. The method according to claim 4, characterized in that The step of performing structural analysis on the standard enterprise file by the structure planning agent to obtain a target structure type of the standard enterprise file includes: Constructing a file structure type library of the target enterprise; wherein the file structure type library stores preset file structure types and structural features of the preset file structure types; Performing structural analysis on the standard enterprise file to obtain file structure features of the standard enterprise file; Based on matching the structural features of the preset file structure type with the file structure features, a target structure type of each of the standard enterprise files is determined from the preset file structure types in the file structure type library.

7. The method according to claim 6, characterized in that The step of performing hierarchical decomposition of the standard enterprise file by the content planning agent to obtain file granularity level data of the standard enterprise file includes: Determining preset granularity data of the target enterprise; wherein the preset granularity data is used to characterize the granularity into which standard enterprise files can be divided; Based on the preset granularity data and the target structure type, file granularity level data of the standard enterprise file is determined.

8. The method according to claim 4, characterized in that The standard enterprise file is obtained by the following method: Extract elements from the initial enterprise documents to obtain initial text elements and initial non-text elements; Performing text conversion on the initial non-text element to obtain a converted text element; The initial text element and the converted text element are saved according to the specified format to obtain the standard enterprise file.

9. A device for storing and querying structured enterprise data applied to an AI vice president system, characterized in that: The enterprise data is structured and stored in an enterprise knowledge base. The device includes: A query task determination module is used to determine enterprise data query tasks targeting the operational behavior of a target enterprise; a department matching module, configured to perform department matching on the first level of the enterprise knowledge base based on the enterprise data query task, and obtain a target department object for executing the operation behavior; wherein the enterprise knowledge base stores enterprise data in a hierarchical structure, the hierarchical structure comprising a first level and a second level, the first level being used to describe the organizational structure of the target enterprise, and the second level being used to describe actual conditions related to the operation behavior; a data matching module, configured to perform data matching on the second level of the enterprise knowledge base based on the enterprise data query task to obtain a target data object generated by the operation behavior; wherein the search scope corresponding to the second level is smaller than the search scope corresponding to the first level; A result output module is used to output the query result corresponding to the enterprise data query task based on the target department object and the target data object.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 8 by executing the computer instructions.

Images