A knowledge graph-based financial audit chain construction method and system

Abstract

This invention discloses a method and system for constructing a financial audit chain based on a knowledge graph, belonging to the field of financial audit technology. It addresses the technical problems of existing enterprise business systems lacking unified data standards and paradigms, leading to cumbersome and error-prone audit processes, and difficulties in locating and easily damaged paper-based original vouchers, all of which hinder financial audit work. The method includes: performing ontology modeling on each business system within the current enterprise to obtain a corresponding financial audit ontology model; collecting financial audit data from each business system and preprocessing the financial audit data; identifying audit entities in the preprocessed financial audit data based on the financial audit ontology model; extracting triplet relationships from the audit entities and constructing a financial domain knowledge graph; and retrieving financial data from the financial domain knowledge graph through a unified query interface, returning the corresponding entities and relationships.

Description

Technical Field

[0001] This invention relates to the field of financial auditing technology, and in particular to a method and system for constructing a financial auditing chain based on knowledge graphs. Background Technology

[0002] As businesses become increasingly complex and diverse, financial auditing is also becoming more challenging. Currently, many business systems operate independently, lacking unified data standards and paradigms. This means that to create a complete audit trail, it's necessary to log into multiple business systems for queries, making the entire audit process cumbersome and prone to errors. Furthermore, the original documents required for auditing are mostly paper-based, making them difficult to review and search, and they are easily damaged, hindering the preservation and retrieval of audit materials. Summary of the Invention

[0003] This invention provides a knowledge graph-based method and system for constructing a financial audit chain to address the following technical problems: the lack of unified data standards and paradigms among existing enterprise business systems leads to cumbersome and error-prone auditing processes, and paper-based original vouchers are difficult to find and easily damaged, which is detrimental to financial auditing work.

[0004] The embodiments of the present invention adopt the following technical solutions:

[0005] On one hand, embodiments of the present invention provide a method for constructing a financial audit chain based on a knowledge graph. The method includes: performing ontology modeling on each business system in the current enterprise to obtain a corresponding financial audit ontology model; collecting financial audit data from each business system and preprocessing the financial audit data; identifying audit entities in the preprocessed financial audit data according to the financial audit ontology model; extracting triple relationships from the audit entities and constructing a financial domain knowledge graph; and retrieving financial data from the financial domain knowledge graph through a unified query interface to return the corresponding entities and relationships.

[0006] In one feasible implementation, ontology modeling is performed on each business system within the current enterprise to obtain a corresponding financial audit ontology model. Specifically, this includes: acquiring the data interaction relationships and business processes between the various business systems of the current enterprise; wherein the business systems at least include a financial system, a contract management system, and a project management system; determining the dependencies between the various business systems based on the data interaction relationships and business processes; determining the business domains to be included in the financial audit chain based on the current enterprise's financial audit requirements; determining entity definitions related to financial audit within the business domains; wherein the entity definitions at least include the entity's domain, entity type, and entity attributes; determining the entity relationships and semantic relationships between the entity definitions based on the dependencies; and creating a financial audit ontology model for each business domain based on the entity definitions, the entity relationships, and the semantic relationships.

[0007] In one feasible implementation, financial audit data is collected from each of the business systems, specifically including: obtaining the sequence of business process nodes for each business system; based on the sequence of business process nodes, automatically triggering a data collection component when an audit process node is reached, so as to dynamically collect the financial audit data of the current business system through a data collection interface; wherein, the financial audit data includes at least semi-structured data and unstructured data.

[0008] In one feasible implementation, the step of dynamically collecting financial audit data from the current business system through a data acquisition interface specifically includes: obtaining a unique identifier for the financial audit data; converting the financial audit data into an XML file based on the XML elements of the business corresponding to the financial audit data; wherein the XML elements include at least: ID, Name, and Label; passing the unique identifier and the XML file to the data receiving interface; for cases where there is an upstream and downstream relationship between the financial audit data, determining a unique identifier pair between two related financial audit data; and converting the relationship into a connection clue between corresponding XML files based on the unique identifier pair.

[0009] In one feasible implementation, the financial audit data is preprocessed, specifically including: performing data value detection on the collected financial audit data; wherein, the data value detection includes at least data integrity detection, accuracy detection, usability detection, and security detection; performing OCR recognition on scanned documents and images in the financial audit data, extracting text information and converting it into a standardized XML file; parsing the XML file, and according to the financial audit ontology model, performing entity definition matching on the XML metadata in the XML file to obtain the corresponding ontology model entities; converting data from different sources into a unified format standard, and standardizing the data storage format.

[0010] In one feasible implementation, based on the financial audit ontology model, identifying audit entities in the preprocessed financial audit data specifically includes: constructing paired components of the XML files and XML metadata according to the relationship between the XML files and XML metadata corresponding to the financial audit data; linking the paired components of multiple XML files into a business relationship knowledge graph based on the association clues between the XML files, and storing it in a structured data format with a master-slave structure; wherein the stored master table contains at least the following information: file ID, file name, file category, and business type, and the slave table contains at least the following information: file ID, file name, and whether it is a master file; extracting XML metadata from the XML nodes; wherein the XML node includes a master file and multiple sub-components; the XML metadata includes at least the sub-component ID, sub-component name, and whether it is a master file; matching the corresponding entity definition in the financial audit ontology model based on the XML metadata, and mapping it to the corresponding audit entity.

[0011] In one feasible implementation, based on the XML metadata, matching the corresponding entity definition in the financial audit ontology model and mapping it to the corresponding audit entity specifically includes: obtaining ontology configuration information in the financial audit ontology model; wherein the ontology configuration information includes at least entity definitions and entity relationship definitions; parsing the XML metadata and identifying entity tag labels therein; traversing the entity definitions in the ontology configuration information and matching them with the entity tag labels, and returning the corresponding entity identifier after matching a corresponding entity definition; in the financial audit ontology model, determining whether the matched entity definition has a corresponding entity relationship definition, and if so, mapping the XML metadata to the specific entity location in the financial audit ontology model through tag attributes; determining whether the current entity is a head entity or a tail entity based on the specific entity location and the description of the subject-verb-object triple relationship in the financial audit ontology model; if the current entity is a head entity, determining the key attribute information of the tail entity based on the entity relationship definition existing in the entity definition matched by the XML metadata, and creating the tail entity.

[0012] In one feasible implementation, triple relationships are extracted from the audit entity, and a financial domain knowledge graph is constructed. Specifically, this includes: constructing triple relationships with the main file as the head entity and the sub-component as the tail entity based on the obtained audit entity; and storing the triple relationships in a graph database through a unified graph data storage API to form the financial domain knowledge graph.

[0013] In one feasible implementation, financial data is retrieved from the financial domain knowledge graph through a unified query interface, returning the corresponding entities and relationships. Specifically, this includes: storing the data in the financial domain knowledge graph in a graphical manner into a graph database through the graph database's API interface; and querying and retrieving knowledge from the graph database through a unified query API interface. The query API interface supports at least traversal queries, filtering queries, and aggregation queries.

[0014] On the other hand, embodiments of the present invention also provide a knowledge graph-based financial audit chain construction system. The system includes: an ontology modeling module for performing ontology modeling on various business systems within the current enterprise to obtain corresponding financial audit ontology models; a data acquisition module for collecting financial audit data from the various business systems and preprocessing the financial audit data; a knowledge graph construction module for identifying audit entities in the preprocessed financial audit data based on the financial audit ontology models; extracting triplet relationships from the audit entities and constructing a financial domain knowledge graph; and a knowledge query module for retrieving financial data from the financial domain knowledge graph through a unified query interface and returning the corresponding entities and relationships.

[0015] Compared with existing technologies, the financial audit chain construction method and system based on knowledge graph provided in this invention have the following beneficial effects:

[0016] (1) Improved audit efficiency and accuracy: Through automated processing and knowledge graph-based retrieval and analysis, key information can be quickly located, reducing the error rate of manual operation and improving audit efficiency and accuracy.

[0017] (2) It integrates various business systems of the enterprise: By building a unified knowledge graph model, the data of various systems of the enterprise can be integrated, breaking the information silo phenomenon and realizing data sharing and collaborative utilization.

[0018] (3) Support audit decisions: By conducting in-depth analysis and mining of knowledge in the knowledge graph, data support and decision-making basis can be provided to enterprises, helping them to make more scientific and reasonable decisions. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0020] Figure 1 A flowchart illustrating a knowledge graph-based financial audit chain construction method provided in this embodiment of the invention;

[0021] Figure 2 A knowledge extraction flowchart provided for an embodiment of the present invention;

[0022] Figure 3 This is a schematic diagram of a knowledge graph-based financial audit chain construction system provided in an embodiment of the present invention. Detailed Implementation

[0023] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0024] This invention provides a method for constructing a financial audit chain based on a knowledge graph, such as... Figure 1 As shown, the method for constructing a financial audit chain based on knowledge graphs specifically includes steps S101-S105:

[0025] S101. Perform ontology modeling on each business system in the current enterprise to obtain the corresponding financial audit ontology model.

[0026] Specifically, the first step is to obtain the data interaction relationships and business processes between the various business systems of the current enterprise; these business systems include the financial system, contract management system, and project management system, etc. Then, based on the data interaction relationships and business processes, the dependencies between the various business systems are determined.

[0027] Furthermore, based on the current financial audit needs of the enterprise, the business areas that need to be included in the financial audit chain are determined. Within these business areas, the entity definitions related to financial audit are determined; these entity definitions must at least include the entity's domain, entity type, and entity attributes.

[0028] Furthermore, based on the dependencies between various business systems, the entity relationships and semantic relationships between the entity definitions are determined; based on the entity definitions, entity relationships, and semantic relationships, a financial audit ontology model for each business domain is created.

[0029] Entity definitions clarify the domain, type, and attributes of an entity. Entity relation definitions clarify the associations and semantic relationships between entities. When constructing a knowledge graph, these entity definitions and entity relation definitions are used to transform the entities into nodes and edges in the knowledge graph.

[0030] As a feasible implementation method, the specific process of constructing an ontology model is as follows: 1. Analyze business systems: Thoroughly understand and analyze the various business systems within the enterprise, including the financial system, contract management system, project management system, etc., then identify the data interactions and business processes between these systems, and understand their dependencies. 2. Determine business domains and identify entities: Based on business needs, determine the business domains that need to be included in the financial audit chain. Ensure the accuracy and completeness of the data, and conduct preliminary quality control on the data. Identify entities related to financial auditing, such as accounting subjects, invoices, contracts, projects, etc.

[0031] S102. Collect financial audit data from various business systems and preprocess the financial audit data.

[0032] Specifically, the sequence of business process nodes for each business system is first obtained. Then, based on the sequence of business process nodes, the data collection component is automatically triggered when an audit process node is reached, so as to dynamically collect the financial audit data of the current business system through the data collection interface; wherein, the financial audit data includes at least semi-structured data and unstructured data.

[0033] As a feasible implementation method, the data acquisition interface configured in the business system dynamically acquires data based on the business process, automatically triggering the acquisition component when a certain process node is reached. Optionally, for electronic documents and other unstructured data that cannot be directly transmitted through the data interface, RPA technology is used for acquisition to achieve automated identification, parsing, and processing of electronic documents.

[0034] In one embodiment, data related to financial auditing is collected from various business systems. The metadata collection standard is as follows: based on the characteristics of enterprise operation and management data, metadata is classified into nine categories, forming a total of nine metadata schemes. Sub-categories of metadata schemes are formed for accounting business-related data, including 91 items of accounting voucher metadata, 42 items of original document metadata, and 20 items of electronic invoice metadata, totaling 153 items. The unstructured data collection standard requires the generation of formatted files (PDF, OFD). Externally received data is directly transmitted in its original form, while data generated by internal systems is converted into formatted files. Structured data is converted using conversion templates, and text files are converted using a unified file format conversion tool.

[0035] Furthermore, financial audit data from the current business system is dynamically collected through a data acquisition interface. This includes: obtaining a unique identifier for the financial audit data; converting the financial audit data into an XML file based on the XML elements corresponding to the business data; wherein the XML elements include at least: ID, Name, and Label. The unique identifier and the XML file are then passed to the data receiving interface.

[0036] Furthermore, for situations where there are upstream and downstream relationships between financial audit data, a unique identifier pair is identified between the two related financial audit data. Then, based on the unique identifier pair, the relationship is converted into a connection thread between corresponding XML files. This solves the problem in existing technologies where business behavior threads originating from multi-source heterogeneous systems are fragmented and difficult to associate with electronic documents.

[0037] As a feasible implementation method, data collected from the business system is converted into semi-structured and unstructured data for storage, while establishing the relationship between the two. Business data is output as a layout file using a virtual printing control. After receiving the layout file, the data acquisition interface returns its unique identifier. Further, the collected business data is converted into an XML file according to the corresponding business's XML Schema (required elements include: ID, Name, Label). The unique identifier returned in the previous step, along with the XML file, is passed to the data receiving interface, establishing a link between the XML file and its XML metadata.

[0038] Furthermore, the preprocessing method for the collected financial audit data specifically includes: performing data value detection on the collected financial audit data; wherein, data value detection includes at least data integrity detection, accuracy detection, usability detection, and security detection. Then, OCR recognition is performed on scanned documents and images in the financial audit data to extract text information and convert it into a standardized XML file. The XML file is parsed, and entity definition matching is performed on the XML metadata in the XML file according to the financial audit ontology model to obtain the corresponding ontology model entities. Finally, data from different sources is converted into a unified format standard, and the data storage format is standardized.

[0039] As a feasible implementation method, data preprocessing includes at least the following steps: (1) Data value detection: Data value detection is performed on the collected raw data, including the detection of data integrity, accuracy, availability, security, etc. Through this step, unreasonable or erroneous information in the data can be discovered and corrected, and the credibility and availability of the data can be improved. In this step, the number, size, timestamp, encryption information, etc. of the XML file will be checked to see if they are consistent with the description in the metadata, and whether the metadata of the XML file meets the data collection specification requirements, etc. (2) Entity detection and parsing of XML files: Entities are identified. If the XML file contains Label bytes and there is matching information in the above entity model, it is mapped to the corresponding ontology model entity. (3) Information extraction: OCR recognition is performed on scanned documents, pictures, etc., to extract text information and convert it into structured data for storage, and convert it into a standardized XML file. ID, Name, and Label information are marked in the XML file, and the entity detection sub-process in this step is called to complete entity matching. (4) Format conversion: Data from different sources and formats are converted into a unified format and standard. The data storage format is standardized according to the collection requirements. Non-standard files are converted into standardized storage formats, while ensuring the availability of data.

[0040] S103. Based on the financial audit ontology model, identify the audit entities in the preprocessed financial audit data.

[0041] Specifically, based on the relationship between the XML files and XML metadata corresponding to the financial audit data, a pairing component for the XML files and XML metadata is constructed. Based on the association clues between the XML files, the pairing components of multiple XML files are linked into a business relationship knowledge graph, and stored in a structured data format with a master-slave structure. The stored master table must contain at least the following information: file ID, file name, file category, and business type; the slave table must contain at least the following information: file ID, file name, and whether it is a master file.

[0042] Furthermore, XML metadata is extracted from the XML node; wherein the XML node includes a main file and multiple child components; the XML metadata includes at least the child component ID, the child component name, and whether it is the main file.

[0043] Further, based on the XML metadata, the corresponding entity definitions are matched in the financial audit ontology model and mapped to the corresponding audit entities. This involves the following steps: obtaining the ontology configuration information in the financial audit ontology model; whereby the ontology configuration information includes at least entity definitions and entity relationship definitions. Then, the XML metadata is parsed, and entity tagging is identified. The entity definitions in the ontology configuration information are traversed and matched against the corresponding entity tag. Upon matching a corresponding entity definition, the corresponding entity identifier is returned. In the financial audit ontology model, it is determined whether the matched entity definition has a corresponding entity relationship definition. If so, the XML metadata is mapped to the specific entity location in the financial audit ontology model through tag attributes. Based on the specific entity location and the description of the subject-verb-object triple relationship in the financial audit ontology model, the current entity is determined to be either a head entity or a tail entity. If the current entity is a head entity, the key attribute information of the tail entity is determined based on the entity relationship definition existing in the entity definition matched from the XML metadata, and the tail entity is created.

[0044] As a feasible implementation method, firstly, based on the relationship between XML files and their metadata, paired components of XML files and their metadata are organized. Then, based on the business relationships between XML files, multiple paired XML file components are linked into a behavioral business relationship knowledge graph, which is stored using a master-slave structured data. The master table contains information such as ID, Name, file category, and business type, while the slave tables include information such as file ID, Name, and whether it is a master file. Standardized XML metadata is then extracted and generated. XML nodes include component ID, sub-component ID, sub-component name, and whether it is a master file. Multiple sub-components generate multiple sections, and each XML has only one master file.

[0045] As a feasible implementation method, Figure 2 A knowledge extraction flowchart is provided for an embodiment of the present invention, such as Figure 2As shown, knowledge extraction includes the following steps: First, a scheduling task is initiated to obtain ontology configuration information, including entity and relation definitions. Then, entity recognition is performed: the XML dataset is retrieved, the XML data is parsed, entity tag labels are identified, entity definitions are traversed, and after matching a specific entity, the corresponding entity identifier is returned. Next, entity mapping of XML metadata is performed: after identifying specific entities, if entity relation mappings exist in the ontology model, the XML metadata is further mapped to specific entities in the knowledge graph ontology model through tag attributes, for subsequent creation of graph database entities. Further, entity location identification is performed on the XML metadata: based on the description of subject-verb-object triple relationships in the ontology model, the current entity position (head or tail entity) is identified. Finally, key attributes of the tail entity are extracted: if the current entity is a head entity, the XML data is parsed, and relation mappings in the ontology model are matched to form the key attribute information of the tail entity, thereby creating the tail entity. Furthermore, if the tail entity is retrieved during future iterations of the dataset, the attribute information of the tail entity is updated.

[0046] S104. Extract triple relationships from audit entities and construct a financial domain knowledge graph.

[0047] Specifically, based on the obtained audit entities, a triple relationship is constructed with the main file as the head entity and the sub-components as the tail entities. This triple relationship is then stored in a graph database using a unified graph data storage API, forming a financial domain knowledge graph.

[0048] As a feasible implementation method, based on the extracted XML metadata, entity recognition is performed according to the component ID and multiple sub-component IDs to form a triple relationship where multiple main files are header entities and other sub-components are tail entities.

[0049] S105. Through a unified query interface, financial data is retrieved in the financial domain knowledge graph, and the corresponding entities and relationships are returned.

[0050] Specifically, data from the financial domain knowledge graph is stored graphically in the graph database through the graph database's API interface. Then, the knowledge in the graph database is queried and retrieved through a unified query API interface; the query API interface supports at least traversal queries, filtering queries, and aggregation queries.

[0051] The constructed knowledge graph enables financial auditing. Through querying and reasoning, potential problems or risks can be quickly identified. Furthermore, this invention provides a visual interface to facilitate auditors' understanding and analysis of financial data.

[0052] In addition, embodiments of the present invention also provide a financial audit chain construction system based on knowledge graphs, such as... Figure 3As shown, the knowledge graph-based financial audit chain construction system 300 specifically includes:

[0053] The ontology modeling module 310 is used to perform ontology modeling on various business systems in the current enterprise to obtain the corresponding financial audit ontology model.

[0054] The data acquisition module 320 is used to collect financial audit data from the various business systems and to preprocess the financial audit data.

[0055] The knowledge graph construction module 330 is used to identify audit entities in the preprocessed financial audit data based on the financial audit ontology model; extract triple relationships from the audit entities; and construct a financial domain knowledge graph.

[0056] The knowledge query module 340 is used to retrieve financial data in the financial domain knowledge graph through a unified query interface and return the corresponding entities and relationships.

[0057] The various embodiments in this invention are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0058] The foregoing has described specific embodiments of the present invention. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0059] The above description is merely an embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the embodiments of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the embodiments of the present invention should be included within the protection scope of the present invention.

Claims

1.A knowledge graph-based financial audit chain construction method, characterized in that, The method includes: Ontology modeling is performed on each business system in the current enterprise to obtain the corresponding financial audit ontology model; wherein, the business systems include at least the financial system, the contract management system, and the project management system; Financial audit data is collected from each of the business systems, and the financial audit data is preprocessed. Based on the aforementioned financial audit ontology model, audit entities are identified in the preprocessed financial audit data, specifically including: Based on the relationship between the XML file and XML metadata corresponding to the financial audit data, construct a pairing component for the XML file and XML metadata; Based on the association clues between XML files, the paired components of multiple XML files are linked into a business relationship knowledge graph, and stored in a structured data format with a master-slave structure. The master table after storage contains at least the following information: file ID, file name, file category, and business type, and the slave table contains at least the following information: file ID, file name, and whether it is a master file. Extract XML metadata from an XML node; wherein the XML node includes a main file and multiple sub-components; the XML metadata includes at least the sub-component ID, the sub-component name, and whether it is the main file; Based on the XML metadata, the corresponding entity definitions are matched in the financial audit ontology model and mapped to the corresponding audit entities, specifically including: Obtain the ontology configuration information in the financial audit ontology model; wherein, the ontology configuration information includes at least entity definitions and entity relationship definitions; Parse the XML metadata and identify the entity tag within it; Iterate through the entity definitions in the ontology configuration information and match them with the entity tag. After matching the corresponding entity definition, return the corresponding entity identifier. In the financial audit ontology model, it is determined whether the matched entity definition has a corresponding entity relationship definition. If so, the XML metadata is mapped to the specific entity location in the financial audit ontology model through tag attributes. Based on the specific entity location and the description of the subject-verb-object triple relationship in the financial audit ontology model, determine whether the current entity is a head entity or a tail entity; If the current entity is a head entity, then based on the entity relationship definition that exists in the entity definition matched by the XML metadata, determine the key attribute information of the tail entity and create the tail entity; Extract triplet relationships from the audit entities and construct a financial domain knowledge graph; Financial data can be retrieved from the financial domain knowledge graph through a unified query interface, returning the corresponding entities and relationships. 2.The knowledge graph-based financial audit chain construction method of claim 1, wherein, Ontology modeling is performed on various business systems within the current enterprise to obtain the corresponding financial audit ontology model, specifically including: To obtain the data interaction relationships and business processes between the various business systems of the current enterprise; Based on the data interaction relationships and business processes, the dependencies between various business systems are determined; Based on the current financial audit needs of the enterprise, determine the business areas that need to be included in the financial audit chain; Within the aforementioned business domain, entity definitions related to financial auditing are determined; wherein, the entity definitions include at least the domain to which the entity belongs, the entity type, and the entity attributes; Based on the dependencies, determine the entity relationships and semantic relationships between the entity definitions; Based on the entity definitions, entity relationships, and semantic relationships, create financial audit ontology models for each business domain. 3.The knowledge graph-based financial audit chain construction method of claim 1, wherein, Collecting financial audit data from the various business systems specifically includes: Obtain the sequence of business process nodes for each business system; Based on the sequence of business process nodes, a data acquisition component is automatically triggered when an audit process node is reached, so as to dynamically collect the financial audit data of the current business system through the data acquisition interface; wherein, the financial audit data includes at least semi-structured data and unstructured data. 4.The knowledge graph-based financial audit chain construction method of claim 3, wherein, The dynamic collection of financial audit data from the current business system via the data acquisition interface specifically includes: A unique identifier for obtaining the financial audit data; Based on the XML elements corresponding to the business transactions in the financial audit data, the financial audit data is converted into an XML file; wherein, the XML elements include at least: ID, Name, and Label; The unique identifier and the XML file are passed to the data receiving interface; When there are upstream and downstream relationships between financial audit data, a unique identifier pair is determined for the two related financial audit data; based on the unique identifier pair, the relationship is converted into a relationship thread between corresponding XML files. 5.The knowledge graph-based financial audit chain construction method of claim 4, wherein, The preprocessing of the financial audit data specifically includes: The collected financial audit data is subjected to data value testing; wherein the data value testing includes at least the data integrity testing, accuracy testing, usability testing, and security testing. OCR recognition is performed on the scanned documents and images in the financial audit data to extract text information and convert it into a standardized XML file. The XML file is parsed, and based on the financial audit ontology model, the entity definition is matched with the XML metadata in the XML file to obtain the corresponding ontology model entity; Transform data from different sources into a unified format standard and standardize the data storage format. 6.The knowledge graph-based financial audit chain construction method of claim 1, wherein, Extracting triplet relationships from the audit entity and constructing a financial domain knowledge graph, specifically including: Based on the obtained audit entities, construct a triplet relationship with the main file as the head entity and the sub-components as the tail entities; The triple relationships are stored in a graph database through a unified graph data storage API to form the financial domain knowledge graph. 7.The knowledge graph-based financial audit chain construction method of claim 1, wherein, Financial data can be retrieved from the financial domain knowledge graph through a unified query interface, returning the corresponding entities and relationships, specifically including: The data in the financial domain knowledge graph is stored in the graph database in a graphical manner through the graph database API interface; The knowledge in the graph database can be queried and retrieved through a unified query API interface; wherein the query API interface supports at least traversal query, filtering query and aggregation query. 8.A knowledge graph based financial audit chain construction system, configured to implement the knowledge graph based financial audit chain construction method according to any one of claims 1-7. The system includes: The ontology modeling module is used to perform ontology modeling on various business systems in the current enterprise to obtain the corresponding financial audit ontology model. The data acquisition module is used to collect financial audit data from the various business systems and to preprocess the financial audit data. The knowledge graph construction module is used to identify audit entities in the preprocessed financial audit data based on the financial audit ontology model; extract triple relationships from the audit entities; and construct a financial domain knowledge graph. The knowledge query module is used to retrieve financial data from the financial domain knowledge graph through a unified query interface and return the corresponding entities and relationships.

Images