Financial report generation method and system based on artificial intelligence

Abstract

The invention relates to the technical field of data processing, and discloses a financial report generation method and system based on artificial intelligence. The method comprises the steps of preprocessing multi-source financial data to obtain target data; performing semantic understanding through a capital extraction algorithm to obtain a knowledge base; intelligent matching template filling is carried out to generate a preliminary report; verifying and analyzing the logic relationship by applying a causal relationship; optimizing the content based on the verification result; and performing multi-level auditing and updating to obtain a final report. According to the method, intelligent processing, deep semantic understanding and logic verification of financial data are realized by introducing a knowledge-based capital extraction algorithm and a causal relationship extraction model, so that a high-quality financial report which is strict in logic and professional in content is generated.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and in particular, to a financial report generation method and system based on artificial intelligence. Background Art

[0002] With the continuous improvement of the enterprise financial reporting system, the compilation of financial reports has become increasingly complex. The traditional compilation of financial reports mainly relies on manual methods. Financial personnel need to extract data from the enterprise financial system, use spreadsheet software for calculation and collation, and manually write the text content of the report. In this way, financial personnel need to invest a large amount of time and energy in data collection, verification, calculation, and report writing. With the advent of the big data era, the volume of enterprise financial data has increased explosively, and the internal financial data, external market data, and industry regulatory data of the enterprise have formed a multi-source heterogeneous data environment, which makes the traditional manual method face problems such as low efficiency and easy errors when processing such a large amount of complex financial data.

[0003] Most of the existing technologies can only process structured financial data, have limited ability to process unstructured financial text information, and cannot deeply understand the semantic information behind the financial data. Secondly, the existing technologies lack the ability to analyze the causal relationship between financial events, making it difficult to conduct logical verification on the content of financial reports, resulting in possible logical contradictions or unreasonable points in the generated reports. Moreover, the existing technologies mostly adopt the template filling method in report content generation, lack the intelligent content generation and optimization ability, and cannot automatically generate personalized and professional report content according to the enterprise characteristics and industry background. In addition, the existing technologies lack a multi-level and multi-dimensional intelligent review mechanism in report review and update, and cannot comprehensively ensure the accuracy and standardization of the report. Summary of the Invention

[0004] This application provides a financial report generation method and system based on artificial intelligence, which is used to realize the intelligent processing, deep semantic understanding, and logical verification of financial data by introducing a knowledge-based capital extraction algorithm and a causal relationship extraction model, so as to generate a high-quality financial report with strict logic and professional content.

[0005] First aspect, the present application provides a method for generating a financial report based on artificial intelligence. The method for generating a financial report based on artificial intelligence includes: collecting and preprocessing multi-source financial data to obtain target financial data; inputting the target financial data into a knowledge-based capital extraction algorithm for financial semantic understanding to obtain a financial data knowledge base; performing intelligent report template matching and automatic financial information filling according to the financial data knowledge base to obtain a first financial report; performing financial logic verification on the first financial report through a causal relationship extraction model, identifying the causal relationship between financial events and comparing with an expected model to obtain a financial logic verification report; generating and optimizing the content of the first financial report through deep learning based on the financial logic verification report to generate a second financial report; performing multi-level automatic review and intelligent update on the second financial report to obtain a target financial report.

[0006] Second aspect, the present application provides a system for generating a financial report based on artificial intelligence. The system for generating a financial report based on artificial intelligence includes:

[0007] A collection module, configured to collect and preprocess multi-source financial data to obtain target financial data;

[0008] An input module, configured to input the target financial data into a knowledge-based capital extraction algorithm for financial semantic understanding to obtain a financial data knowledge base;

[0009] A filling module, configured to perform intelligent report template matching and automatic financial information filling according to the financial data knowledge base to obtain a first financial report;

[0010] A verification module, configured to perform financial logic verification on the first financial report through a causal relationship extraction model, identify the causal relationship between financial events and compare with an expected model to obtain a financial logic verification report;

[0011] A generation module, configured to generate and optimize the content of the first financial report through deep learning based on the financial logic verification report to generate a second financial report;

[0012] An update module, configured to perform multi-level automatic review and intelligent update on the second financial report to obtain a target financial report.

[0013] Third aspect, there is provided a device for generating a financial report based on artificial intelligence, including: a memory and at least one processor, wherein instructions are stored in the memory; the at least one processor invokes the instructions in the memory so that the device for generating a financial report based on artificial intelligence executes the above-mentioned method for generating a financial report based on artificial intelligence.

[0014] Fourthly, a computer-readable storage medium is provided, in which instructions are stored. When the instructions run on a computer, the computer is enabled to execute the above-mentioned financial report generation method based on artificial intelligence.

[0015] In the technical solution provided by this application, through multi-source financial data collection and preprocessing, the problems of scattered data sources and inconsistent formats in the traditional financial report compilation process are solved, and the data processing efficiency is significantly improved; the knowledge-based capital extraction algorithm adopted breaks through the limitation of the traditional method's superficial understanding of financial data, realizes the intelligent parsing of the deep semantics of financial data, accurately identifies the intellectual capital elements hidden in the financial data, and enables the financial report to comprehensively reflect the enterprise value; the intelligent report template matching and automatic filling technology overcomes the rigidity of the traditional template usage, dynamically selects and adjusts the optimal template according to the enterprise characteristics and the characteristics of financial data, and realizes the personalized customization of the report; the application of the causal relationship extraction model is the core innovation point of the present invention. By integrating Chinese language characteristics and graph attention network technology, this model accurately identifies the causal logic relationship between financial events, effectively solves the problems of fuzzy causal relationships and logical contradictions in traditional financial reports, and makes the report content logically rigorous and well-organized; the content generation and optimization technology based on deep learning breaks through the limitations of single content and mechanical expression in traditional report generation methods, and intelligently generates professional and accurate explanatory texts according to the characteristics of financial data and verification results, enhancing the readability and professional depth of the report; the multi-level automatic review and intelligent update technology constructs an all-round quality assurance system. Through multi-dimensional detections such as cross-verification, trend analysis, and semantic matching, it ensures the accuracy, integrity, and standardization of the report, greatly reduces the report error rate, improves the generation efficiency of the financial report, and more importantly, guarantees the quality and professional level of the report, providing an intelligent solution for the entire process of enterprise financial report compilation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a schematic diagram of an embodiment of the financial report generation method based on artificial intelligence in an embodiment of this application;

[0018] Figure 2 It is a schematic diagram of an embodiment of the financial report generation system based on artificial intelligence in an embodiment of this application;

[0019] Figure 3It is a schematic block diagram of a financial report generation device based on artificial intelligence in an embodiment of the present invention. Detailed implementation manners

[0020] Embodiments of the present application provide a financial report generation method and system based on artificial intelligence. Terms such as first, second, third, fourth, etc. (if any) in the specification, claims, and above-mentioned drawings of the present application are used to distinguish similar objects and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments described herein can be implemented in an order other than those illustrated or described herein. In addition, the term "including" or "having" and any deformation thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0021] For ease of understanding, the following describes the specific process of the embodiments of the present application. Please refer to Figure 1 An embodiment of the financial report generation method based on artificial intelligence in the embodiments of the present application includes:

[0022] Step S101: Collect and preprocess multi-source financial data to obtain target financial data;

[0023] Step S102: Input the target financial data into a knowledge-based capital extraction algorithm for financial semantic understanding to obtain a financial data knowledge base;

[0024] Step S103: Perform intelligent report template matching and automatic financial information filling based on the financial data knowledge base to obtain a first financial report;

[0025] Step S104: Verify the financial logic of the first financial report through a causal relationship extraction model, identify the causal relationship between financial events and compare it with the expected model to obtain a financial logic verification report;

[0026] Step S105: Generate and optimize the content of the first financial report through deep learning based on the financial logic verification report to generate a second financial report;

[0027] Step S106: Perform multi-level automatic review and intelligent update on the second financial report to obtain the target financial report.

[0028] It can be understood that the execution entity of the present application can be a financial report generation system based on artificial intelligence, or a terminal or a server. Specifically, it is not limited here. The embodiments of the present application are described by taking the server as the execution entity as an example.

[0029] Specifically, perform timed data collection from multiple data sources, including the core financial system within the enterprise, external audit report libraries, historical financial statements, and external data from regulatory agencies. These multi-source data often have inconsistent formats and complex content, so they need to undergo format conversion processing to uniformly convert data from different sources into a standardized structure to ensure the smooth progress of subsequent processing. Data conversion not only includes simple format replacement but also requires data integrity checks to identify the absence of key financial fields and fill them according to industry standards to ensure the accuracy and comprehensiveness of the data. After data filling is completed, consistency verification is carried out to check the relationships between major financial statements, such as the balance sheet, income statement, and cash flow statement, to ensure their numerical consistency and logical rationality. For the processing of sensitive data, desensitization processing is carried out, and sensitive information such as customer identifiers and employee salaries is desensitized at different levels to ensure data security. By establishing a financial indicator correlation graph, the system obtains target financial data, laying a solid foundation for subsequent steps.

[0030] Input these preprocessed target financial data into the knowledge-based capital extraction algorithm for financial semantic understanding. The knowledge-based capital extraction algorithm (KBICE) deeply analyzes the semantic content of the data by leveraging technologies such as deep learning, natural language processing, and knowledge graphs. During this process, the system first identifies key terms and numerical information in the financial data, extracts financial keywords, amount values, and relevant time information through a financial professional dictionary and named entity recognition technology. These information are then mapped to a standard financial concept system, enabling the data to be semantically analyzed at a higher level. Through the ontology mapping module, financial terms are transformed into structured financial entities, and syntactic dependency analysis technology is used to identify the logical relationships and semantic structures in the financial data. On this basis, the system also performs event clustering, organizing financial matters with semantic similarity and temporal correlation into event units to construct a financial event sequence. Further, these events are classified and quantified to generate a financial data knowledge base containing rich semantic information, providing strong data support for the generation of intelligent reports.

[0031] Based on the obtained financial data knowledge base, start the intelligent report template matching and automatic filling of financial information. First, extract the financial characteristic data of the enterprise, including asset scale, liability structure, profitability, etc., as the basis for selecting the most suitable report template. By calculating the matching degree between each template in the template library and the financial characteristics of the enterprise, the financial report template that best meets the current needs can be selected. After that, conduct a component evaluation on the selected report template, select the table components suitable for the current enterprise data, and build a customized report structure that conforms to the financial characteristics of the enterprise by combining these components. Then, map the data in the financial data knowledge base to the fields in the report template to clarify the filling rules for each field. Based on these rules, automatically fill the financial data into the report template, and calculate the financial indicators and fill the calculation results into the corresponding positions. The report will be processed through format standardization, adjusting the numerical format and table style to ensure that the report complies with professional norms, and obtain the first financial report. The first financial report will be verified for financial logic through a causal relationship extraction model. This model uses technologies such as graph neural networks and dependency analysis to identify the causal relationships between different financial events in the report. By analyzing the logical order and correlation of financial events, it is possible to compare with the preset causal knowledge base, detect any abnormal causal relationships that do not conform to financial laws, and mark these anomalies. It can reveal potential logical loopholes in the report and ensure the internal consistency and rationality of the report. The system generates a financial logic verification report, which details the financial chains that have passed the verification and the logical anomalies that need to be corrected, and provides correction suggestions.

[0032] Based on the financial logic verification report, conduct deep learning content generation and optimization on the first financial report. This process includes correcting the logical defects in the report, replacing the unreasonable financial logic areas, supplementing the correction basis and conducting text optimization. At the same time, the system will also analyze the changes in key financial indicators, generate professional explanations for the reasons for the fluctuations of the indicators, and enrich the content of the report. It can transform the content in the report into more standardized and professional text, forming the second optimized financial report. The second financial report will be subject to multi-level automatic review and intelligent update. Bidirectional cross-review will be carried out to reverse verify the consistency between the overall financial indicators and the segment data, and at the same time compare with the historical financial report data to identify any abnormal trend changes. The system will also conduct semantic framework analysis to ensure the semantic matching degree between the financial narrative and the corresponding data, and mark the inconsistent parts. Through these review processes, a problem space map can be generated, showing all the detected problem points and providing corresponding correction plans. The report will be updated regionally and directionally according to these correction plans to ensure that the content of the report is more consistent and accurate, and finally generate the target financial report.

[0033] In the embodiments of the present application, through multi-source financial data collection and preprocessing, the problems of scattered data sources and inconsistent formats in the traditional financial report compilation process are solved, significantly improving the data processing efficiency; the knowledge-based capital extraction algorithm adopted breaks through the limitations of the traditional method's superficial understanding of financial data, realizes the intelligent parsing of the deep semantics of financial data, accurately identifies the intellectual capital elements hidden in the financial data, and enables the financial report to comprehensively reflect the enterprise value; the intelligent report template matching and automatic filling technology overcomes the rigidity of traditional template usage, dynamically selects and adjusts the optimal template according to the enterprise characteristics and financial data characteristics, and realizes the personalized customization of the report; the application of the causal relationship extraction model is the core innovation of the present invention. By integrating Chinese language characteristics and graph attention network technology, this model accurately identifies the causal logic relationship between financial events, effectively solves the problems of fuzzy causal relationships and logical contradictions in traditional financial reports, and makes the report content logically rigorous and well-organized; the content generation and optimization technology based on deep learning breaks through the limitations of single content and mechanical expression in traditional report generation methods, intelligently generates professional and accurate explanatory texts according to the financial data characteristics and verification results, enhancing the readability and professional depth of the report; the multi-level automatic review and intelligent update technology constructs an all-round quality assurance system. Through multi-dimensional detections such as cross-verification, trend analysis, and semantic matching, it ensures the accuracy, integrity, and standardization of the report, greatly reduces the report error rate, improves the generation efficiency of the financial report, and more importantly, guarantees the quality and professional level of the report, providing an intelligent solution for the entire process of enterprise financial report compilation.

[0034] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0035] Regularly collect the enterprise's core financial system, audit report library, historical financial statements, and external regulatory data to obtain the original financial data set;

[0036] Perform format conversion processing on the original financial data set, convert the data in PDF reports, spreadsheets, and text documents into a standardized structure to obtain the initial financial data;

[0037] Check the data integrity of the initial financial data, identify the missing items in the key financial fields, and fill them according to industry standards to obtain the complete financial data;

[0038] Perform consistency verification on the complete financial data, check the reconciliation relationship between the balance sheet, income statement, and cash flow statement, and mark the data items that do not conform to accounting standards to obtain the verified financial data;

[0039] Perform sensitive information processing on the verified financial data, classify and desensitize the customer identification and employee salary details to obtain the desensitized financial data;

[0040] Construct a financial indicator correlation graph based on desensitized financial data to obtain target financial data.

[0041] Specifically, regularly collect the enterprise's core financial system, audit report library, historical financial statements, and external regulatory data to obtain the original financial data set. The core of this process lies in using multi-source data collection technology to automatically obtain the enterprise's financial data from various data sources. The enterprise's core financial system usually stores the enterprise's daily financial transaction records, the audit report library provides the audit situation of past financial data, the historical financial statements are detailed records of the enterprise's previous financial conditions, and the external regulatory data may include financial norms from government agencies or industry regulatory departments, industry financial trends, etc. To ensure the timeliness and accuracy of the data, the system needs to regularly obtain data from these sources to ensure that the data used in the financial report preparation process is the latest and most comprehensive. This process often involves interfacing with these data sources and realizing automatic collection through APIs or file uploads, enabling the automatic acquisition of various financial data sets.

[0042] Perform format conversion processing on the original financial data set. The original data usually exists in various formats such as PDF reports, spreadsheets, text documents, etc., and data in different formats need to be converted into a standardized data structure through a unified conversion process. This process often involves parsing different format files and extracting the financial data therein. The tables in PDF reports, the cell data in spreadsheets, and the descriptive information in text documents all need to be processed through data extraction technology. For example, PDF reports usually need to use OCR (Optical Character Recognition) technology to identify the numbers and text content therein, spreadsheets read the values and attributes of each cell, and the financial information in text documents may need to use natural language processing technology to extract the financial keywords. Through this conversion processing, the originally non-uniformly formatted data can be converted into a standardized structure, making subsequent data processing more convenient and efficient.

[0043] After completing the format conversion, data integrity checks are performed on the initial financial data. The integrity of financial data is crucial for the accuracy of reports. Therefore, missing items in key financial fields are automatically identified and filled. For example, the total assets and total liabilities in the balance sheet may be missing, or the values of certain accounts in historical statements are not filled. These missing items will be filled according to industry standards and financial guidelines. The filling process usually estimates based on the financial data of similar enterprises, historical data, and industry trends, and automatically generates reasonable filling values through algorithms to ensure the integrity and consistency of financial data. The filling algorithms and rules used in this process usually rely on industry standards and financial accounting norms. For example, the proportional relationship between different accounts is determined according to accounting standards, or the reasonable value of missing data is inferred based on the growth trend of historical data. After completing the data integrity checks, consistency verification is performed to check the cross-check relationships between different financial statements. The cross-check relationships of financial statements refer to the mutual relationships between the balance sheet, income statement, cash flow statement, etc. For example, the total assets in the balance sheet should be equal to the sum of liabilities and owner's equity, and the net profit in the income statement should match the operating cash flow in the cash flow statement. The system automatically checks the logical consistency between these financial statements through algorithms and marks the data items that do not conform to accounting standards. This process ensures the accuracy and compliance of financial data and avoids data errors and inconsistencies in reports. For example, if the total assets in the balance sheet are inconsistent with the sum of liabilities and owner's equity, this data item will be automatically marked as incorrect, indicating that further correction is required. After the data consistency verification is completed, sensitive information processing is performed on the verified financial data. Sensitive information, such as the personal identification of customers and the salary details of employees, must be desensitized to ensure data security. Desensitization usually includes encrypting or replacing sensitive information to prevent data leakage. For example, the system may replace the customer's name and contact information with symbols, or encrypt the specific salary amount of employees, only retaining the salary range or category information. This enables subsequent financial analysis and report generation while ensuring data privacy and avoiding the leakage of sensitive data. Based on the desensitized financial data, the system constructs a financial indicator correlation graph, laying the foundation for the generation of target financial data. The financial indicator correlation graph is a graph that contains various financial data items and their mutual relationships. The system analyzes the correlation between different financial data through algorithms and establishes a structured financial data graph based on these relationships. For example, there is a certain relationship between current assets and current liabilities in the balance sheet, and there is also a correlation between operating income and net profit in the income statement. Through this graph, the internal connections between different financial data can be intuitively displayed.

[0044] In a specific embodiment, the process of performing step S102 may specifically include the following steps:

[0045] The target financial data is input into the term recognition module in the knowledge-based capital extraction algorithm for processing. Through the financial professional dictionary and named entity recognition technology, financial keywords, amount values, and time information are extracted to obtain a set of financial terms.

[0046] The set of financial terms is input into the ontology mapping module in the knowledge-based capital extraction algorithm for processing. Using the preset knowledge ontology in the financial field, the terms are mapped to the standard financial concept system to obtain structured financial entities.

[0047] The structured financial entities are input into the relationship extraction module in the knowledge-based capital extraction algorithm for financial semantic association analysis. Through syntactic dependency analysis, the logical relationship between the semantic subject and object is identified to obtain a financial event relationship diagram.

[0048] The financial event relationship diagram is input into the event clustering module in the knowledge-based capital extraction algorithm for processing. According to the financial semantic similarity and temporal correlation, the associated events are organized into event units to obtain a financial event sequence.

[0049] The financial event sequence is input into the capital classification module in the knowledge-based capital extraction algorithm for intellectual capital classification. According to the predefined capital category standard, the financial events are classified and quantified to obtain a classified financial knowledge structure.

[0050] The classified financial knowledge structure is input into the weight calculation module in the knowledge-based capital extraction algorithm for processing. By calculating the importance weights of each knowledge node and constructing a multi-level index structure, a financial data knowledge base is obtained.

[0051] Specifically, the target financial data is input into the term recognition module in the knowledge-based capital extraction algorithm. This module extracts key information from the financial data through the financial professional dictionary and named entity recognition (NER) technology. The financial data contains a large number of professional terms, amount values, time information, etc. The task of the term recognition module is to identify these key elements from the original data. For example, in a financial statement, the term recognition module may identify terms such as accounts receivable, total assets, net profit, the first quarter of 2023, etc. Amount values such as 10,000,000 yuan will also be extracted, and time information such as December 31, 2023 will also be recognized and marked. Through the financial professional dictionary, various financial terms and values can be accurately identified, forming a set containing various financial terms.

[0052] The set of financial terms will be input into the ontology mapping module in the knowledge-based capital extraction algorithm. The process of ontology mapping is to map the identified terms to the standard knowledge system in the financial field to ensure the consistency and standardization of the terms. The knowledge ontology in the financial field is usually a preset standard framework that defines the meanings of financial terms and their relationships. Through the ontology mapping module, terms such as accounts receivable will be transformed into a unified representation in the standard financial concept system. For example, accounts receivable may be mapped to Accounts Receivable and defined as the amount that an enterprise collects from customers for selling goods or providing services. This mapping process gives financial terms a unified semantics, enabling in-depth understanding of these terms through the ontology knowledge base, eliminating ambiguities brought by different data sources and formats, and providing structured financial entities for further analysis. After completing the ontology mapping, the financial data will be input into the relationship extraction module in the knowledge-based capital extraction algorithm. This module uses syntactic dependency analysis technology to identify the semantic relationships between different entities in the financial data and establish the logical connections of financial events. For example, in the balance sheet, there is a certain corresponding relationship between assets and liabilities, and there is also a close connection between operating income and net profit in the income statement. Through dependency analysis, these relationships can be identified, such as the balance relationship between assets and liabilities, the causal relationship between income and profit, etc., and represented in a graph structure to form a financial event relationship graph. This graph structure provides a basis for subsequent financial event clustering. It shows the relevance between financial events, helping the system understand the logical connections and influence relationships between different financial data items.

[0053] The financial event relationship graph will be input into the event clustering module in the knowledge-based capital extraction algorithm for processing. At this stage, related events are organized and clustered according to the semantic similarity and temporal correlation of financial semantics. For example, in financial statements, some events such as revenue growth and profit increase may have semantic similarities, while other events such as capital expenditure and debt repayment may be temporally related. The event clustering module will classify these related events according to semantic similarity and temporal correlation to form a financial event sequence. For example, if a company's operating income has increased at multiple time points, these growth events will be classified as a series of revenue growth events and arranged in chronological order to form an ordered financial event sequence. It is possible to extract an ordered event sequence from complex financial data.

[0054] The financial event sequence classifies intellectual capital in the capital classification module of the knowledge-based capital extraction algorithm. The financial events are classified and quantified according to predefined capital category criteria. The classification basis of financial events may include financial elements such as the assets, liabilities, and revenues of an enterprise. For example, the system can classify all financial events related to assets into the capital category and classify events related to liabilities as debt capital. Through this classification, the financial data can be further refined to form a hierarchical financial knowledge structure. This structured financial knowledge helps the system quickly locate relevant data and information when generating reports and provides more accurate support during the analysis process.

[0055] The classified financial knowledge structure will be input into the weight calculation module of the knowledge-based capital extraction algorithm. The role of the weight calculation module is to calculate the importance weights of each knowledge node and construct a multi-level index structure. Each financial event or financial indicator has a different association strength with other events or indicators, and these strengths are quantified by the weight calculation module according to a certain algorithm. The system can evaluate the relative importance of each financial data item in the report and adjust the priority of subsequent report generation based on its importance. For example, if a certain financial data has changed significantly in the past few reporting cycles, the system may assign a higher weight to this data to highlight it in the report. By establishing a multi-level index structure, it is possible to retrieve and call relevant data more efficiently, improving the efficiency and accuracy of report generation.

[0056] In a specific embodiment, the process of executing step S103 may specifically include the following steps:

[0057] Extract data on the enterprise's asset scale, liability structure, and profitability indicators from the financial data knowledge base to obtain the enterprise's financial characteristic data;

[0058] Calculate the template matching degree in the template library based on the enterprise's financial characteristic data, and select the template with the highest matching degree to obtain the preferred report template;

[0059] Conduct a component evaluation on the preferred report template, select table components suitable for the enterprise's data and combine them to obtain a customized report structure;

[0060] Establish a mapping relationship between the fields of the customized report structure and the data nodes in the financial data knowledge base to obtain the field filling rules;

[0061] Fill the data in the financial data knowledge base into the customized report structure according to the field filling rules, calculate the financial indicators and fill in the results to obtain a complete report;

[0062] Perform format standardization processing on the complete report, adjust the numerical format and table style, and mark the areas that need to be supplemented and explained to obtain the first financial report.

[0063] Specifically, the financial data knowledge base is processed to extract the financial characteristic data of the enterprise. These characteristic data mainly include key financial information such as the enterprise's asset scale, liability structure, and profitability indicators. After the data in the financial data knowledge base have undergone preliminary standardization and semantic understanding, they already possess structured financial information. The system can accurately obtain the financial status of the enterprise by extracting this information. For example, the asset scale can be calculated through the asset items listed in the balance sheet, the liability structure can be obtained by analyzing the liability accounts, and profitability indicators such as net profit and gross profit margin can be calculated through the data in the income statement. These extracted financial characteristic data provide key data support for subsequent report generation. After obtaining the financial characteristic data of the enterprise, the system will then match these data with various report templates in the template library and calculate the matching degree of each template. The template library contains a variety of financial report templates, and for enterprises in different industries and of different scales, the structure, content, and format of each template are different. By comparing the financial characteristic data of the enterprise with the structural requirements in the template, the system can evaluate the adaptability of each template. Specifically, information such as the enterprise's asset scale, liability structure, and profitability indicators will be used as inputs, and the system calculates the matching degree of each template with the enterprise's financial data through a preset algorithm, and finally selects the report template that best matches the enterprise's financial data. For example, if the enterprise has a large asset scale, the system may select a template that includes a detailed balance sheet, and if the enterprise's profitability indicators are outstanding, the system may select a template that pays more attention to financial analysis and trend display. Through this matching, it can be ensured that the selected template conforms to the financial characteristics of the enterprise, thus providing a more accurate report framework.

[0064] After selecting the optimal report template, the system enters the component evaluation phase. The content in the template usually consists of multiple table components, which may include the balance sheet, income statement, cash flow statement, etc. In this step, each table component is evaluated to determine whether it is suitable for the current enterprise's needs based on the characteristics of the enterprise's financial data. For example, if the enterprise has a complex debt structure, the system may select a more detailed balance sheet component to display the composition and changes of debts; if the enterprise has a relatively simple asset structure, the system may choose a concise balance sheet component to show the enterprise's asset scale. The selection of components is not only an adjustment of the template, but also a customization of the report structure according to the financial characteristics of the enterprise, making the report content more in line with the actual needs. By combining different table components, a customized report structure can be generated to ensure that the report content covers all aspects of the enterprise's financial situation. The fields of the customized report structure will be mapped to the data nodes in the financial data knowledge base to obtain the field filling rules. Each data item in the financial data knowledge base has its corresponding field position, and each field in the report template also needs to be filled with the corresponding data. In this process, the system ensures that each piece of financial data can be accurately filled into the corresponding report field by establishing the mapping relationship between the data nodes and the fields. For example, the total assets field in the balance sheet should be filled with the asset data from the enterprise's balance sheet, and the net profit field in the income statement should be filled with the net profit data extracted from the income statement. Through these field mapping relationships, the filling process of the financial report can be ensured to be both accurate and efficient.

[0065] Once the field filling rules are clear, the system can fill the data in the financial data knowledge base into the customized report structure according to these rules, calculate financial indicators, and finally generate a complete financial report. At this stage, the system not only fills the original data into the designated positions in the report, but also needs to calculate some key financial indicators, such as net profit, current ratio, gross profit margin, etc. These financial indicators are usually obtained through a series of calculation formulas. For example, net profit can be obtained by subtracting costs and expenses from revenue, and the current ratio can be obtained by dividing current assets by current liabilities. The system will perform necessary calculations on the financial data according to the preset calculation rules and automatically fill the calculation results into the corresponding positions in the report. After these operations, a complete report containing all necessary financial data and indicators will be obtained. The format of the generated complete report is standardized, adjusting the numerical format and table style to ensure that the report meets professional standards. For example, the display format of numbers will be unified to ensure that all amount items are displayed in a unified currency unit and the number of decimal places is consistent; for the data in the table, the typesetting and column width of the table will be adjusted according to the preset style to ensure that the visual effect of the report is clear and tidy. In addition, the system will also mark the areas that need supplementary explanations. For example, some important financial data may require additional notes or explanations, and blank spaces will be left in these areas and marked to prompt the user to make manual supplements. After these format adjustments, the system generates the first financial report that meets the standards and is ready to enter the subsequent review and optimization stage.

[0066] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0067] Decompose the first financial report into multiple financial event units, extract key financial indicator changes and business activities through word segmentation, syntactic parsing, and named entity recognition to obtain a financial event set;

[0068] Construct a financial event graph for the financial event set, take each financial event as a node, and initialize the potential connection relationship between nodes to obtain an initial financial event network;

[0069] Input the initial financial event network into the Chinese dictionary enhancement layer in the causal relationship extraction model, and perform semantic enhancement on the event description through the financial professional dictionary to obtain a semantically enhanced event representation;

[0070] Perform financial event dependency analysis on the semantically enhanced event representation through the graph attention network layer in the causal relationship extraction model, calculate the multi-head attention weights between nodes, identify the causal association strength between events, and obtain a weighted causal relationship graph;

[0071] Compare the weighted causal relationship graph with the preset causal knowledge base in the financial field, detect abnormal causal links and logical contradictions, mark the causal relationships that do not conform to financial laws, and obtain financial logic anomaly marks;

[0072] Generate a financial logic verification report based on the financial logic anomaly markers, including the causal chains that pass the verification and the abnormal causal relationships that need to be corrected, and provide correction suggestions for each anomaly to obtain the financial logic verification report.

[0073] Specifically, decompose the first financial report into multiple financial event units. Through techniques such as word segmentation, syntactic parsing, and named entity recognition (NER) in natural language processing technology, extract the key financial indicator changes and business activities in the report. The purpose of this process is to regard each financial change, each business activity, etc. in the report as an independent financial event, forming a set of financial events. For example, assume that the report mentions that the net profit of the company increased by 10% in the first quarter of 2023. This change will be extracted as an independent event unit and labeled as net profit increase. Through word segmentation technology, the system can accurately segment sentences and identify relevant financial terms, numerical changes, and time nodes. In addition, named entity recognition technology helps the system identify the key financial terms in the event, such as net profit, increase, and the relevant time information of the first quarter of 2023, ensuring the accurate extraction of financial events. These financial event units form a set. After obtaining the set of financial events, the system then constructs a financial event graph, taking each financial event as a node and initializing the potential connection relationships between the nodes. The financial event graph is a graph structure, where each node represents a financial event, and the connections between the nodes represent the potential relationships between the events. For example, if a company's report mentions two events: revenue increase and cost decrease, the system may infer a causal relationship or a chronological relationship between them through business logic and initialize the edge between these two event nodes. The financial event graph provides a graph structure basis for subsequent causal relationship analysis and can visually display the associations between different financial events.

[0074] The financial event graph is input into the causal relationship extraction model for processing. The model first enhances the semantic representation of the event description through the Chinese dictionary enhancement layer. This process improves the understanding of financial terms through a financial professional dictionary. For example, for the event of a 10% increase in net profit, the system identifies the semantic relationship between net profit and a 10% increase through the financial dictionary, enhancing the semantic representation of the event and ensuring the understanding of the financial background and meaning of this change. This enhanced event representation is further processed. Through syntactic dependency analysis and graph neural network (GNN) technology, the system can deeply analyze the dependency relationships between events and identify the causal relationships and semantic associations between events based on the syntactic structure.

[0075] After dependency parsing, the graph attention network (GAT) is used to deeply analyze the financial event graph, calculate the multi-head attention weights between nodes, and identify the causal association strength between events. The graph attention network evaluates the association strength between each pair of nodes by calculating their attention. For example, if there is a strong causal link between the revenue growth event and the net profit growth event, the weight value between them will be large, while for some events with weak associations, such as management changes and revenue growth, the weight may be small. It is able to identify the causal relationship strength between financial events and sort the events according to the weight values, providing a basis for subsequent financial logic verification.

[0076] The causal relationship graph with weights is compared with the pre-set causal knowledge base in the financial domain to detect whether there are abnormal causal links and logical contradictions. For example, the causal knowledge base in the financial domain may contain standard causal rules, such as revenue growth usually leads to net profit growth. If there is a connection in a certain financial event graph where revenue growth leads to a decrease in net profit, it indicates that there may be a logical error. It will automatically compare these causal relationships, mark the causal links that do not conform to financial laws, and form financial logic anomaly marks. These marks help the system identify potential abnormal financial logics in the report and ensure that the generated report is logically self-consistent.

[0077] Based on these financial logic anomaly marks, a financial logic verification report will be generated. The report will include the causal chains that pass the verification and list in detail the abnormal causal relationships that need to be corrected. For each abnormal causal relationship in the report, a correction suggestion will be attached to help reviewers identify and correct potential errors. For example, assuming there is a causal relationship in a report where revenue growth leads to a decrease in net profit, this anomaly will be marked in the financial logic verification report and a correction will be recommended. Possible correction methods may include re-evaluating the relationship between revenue and costs or reviewing the calculation method. The system can not only detect potential problems in the financial report but also provide specific suggestions for correcting the report.

[0078] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0079] Map the financial logic verification report with the first financial report, extract the position marks and corresponding correction suggestions for each area to be corrected, and obtain a correction instruction data set;

[0080] Sort the correction instruction data set according to the quantification value of the anomaly degree, calculate the priority score for each correction task, and obtain a weighted correction task sequence;

[0081] Based on the weighted correction task sequence, perform content reconstruction on the first financial report, replace the text in the unreasonable financial logic area, and supplement the correction basis description to obtain a draft of the logic correction report;

[0082] Generate explanations for key indicators in the draft logical correction report, analyze the reasons for abnormal fluctuations in financial indicators, and add text for professional background analysis to obtain an enhanced report on indicator explanations;

[0083] Convert the text expression of the enhanced report on indicator explanations to obtain a text normalization report;

[0084] Adjust the chapter organization of the text normalization report to obtain the second financial report.

[0085] Specifically, the process of mapping the financial logic verification report with the first financial report and extracting the correction instruction dataset can effectively identify and correct potential problems in the financial report. During this process, the financial logic verification report marks all areas that need to be corrected through comparison and inspection. By comparing the financial events and logic verification results in the report, areas that need to be corrected can be extracted from the logic verification report, including causal relationships that do not conform to financial rules, logical anomalies, and numerical errors, etc. Each correction area will be marked as an area to be corrected and accompanied by corresponding correction suggestions. The correction suggestions may be to adjust a certain financial indicator or value, or to modify the causal relationship. The position markings of these correction areas and the corresponding correction suggestions form the correction instruction dataset. The correction instruction dataset will be sorted according to the quantified value of the degree of abnormality, and the priority of each correction task will be calculated. The calculation basis of the quantified value includes the impact scope of the correction area, the weight of the financial indicator, the resources and time required for correction, etc. For example, for a major error that may cause changes in the overall financial situation of the enterprise (such as a large deviation in net profit), the correction priority will be set to a higher level; while for an error with less impact (such as a small change in accounts receivable), the correction priority will be lower. This process is actually completed based on the hierarchical structure and impact analysis of financial data, ensuring that high-impact errors are processed first and low-impact errors are adjusted subsequently. In the calculation of the priority of correction tasks, the weight relationship of financial data will be used for sorting, and indicators with higher weights will be given higher correction priorities to ensure that the most critical financial problems are solved as early as possible.

[0086] Once the sorting of the correction tasks is completed, the system enters the content reconstruction phase. Based on the weighted sequence of correction tasks, the first financial report will undergo item-by-item content reconstruction, replacing the text in unreasonable financial logic areas and supplementing the description of the correction basis. For example, if an unreasonable logic such as revenue growth leading to a decline in net profit is mentioned in the report, this logical error will be marked in the report and replaced with a more financially regular description, such as revenue growth and cost control jointly leading to an increase in net profit. At the same time, the corresponding correction basis will be supplemented for each corrected area, such as supplementing financial analysis data, industry standards, or accounting standards as support to ensure the accuracy and rationality of the report logic. After the content reconstruction of the first financial report is completed, the system enters the key indicator explanation generation phase. According to the key indicators involved in the financial report, the reasons for abnormal fluctuations are analyzed. For example, if the net profit of an enterprise shows abnormal fluctuations during the reporting period, in-depth analysis will be carried out based on the change trends in the financial data and other relevant data (such as revenue, cost, taxes, etc.) to identify the root causes of the fluctuations, such as rising costs, declining revenue, or the impact of external economic factors. Professional background analysis text will be automatically generated based on these analysis results to explain in detail the reasons for the fluctuations of these financial indicators and supplement them to the report. This process not only provides accurate financial analysis but also helps management understand the business activities behind the financial data, enhancing the practical value of the report.

[0087] The text expression of the enhanced report on indicator explanations will be converted to conform to the standardized report format. The text expression conversion mainly includes the unification of language style, the refinement of expression methods, and the standardization of terms. The language in the report will be adjusted to be more concise and clear, avoiding verbose expressions, and ensuring the report has a high degree of readability and professionalism. The standardization of terms is also crucial. The use of financial terms will be unified to ensure that all financial indicators, accounts, and terms conform to the latest accounting standards and industry norms. The goal of this phase is to ensure that the content of the report is not only accurate and detailed but also highly standardized, conforming to the International Financial Reporting Standards (IFRS) or other relevant financial norms.

[0088] The chapter organization of the text standardized report will be adjusted to ensure that the structure of the report is reasonable and well-organized. This process involves dividing the report into different chapters according to the logic and content of the financial report and optimizing each chapter to ensure that the report has a clear hierarchical structure. For example, a financial report may include chapters such as financial overview, financial analysis, risk assessment, and future outlook. Appropriate adjustments will be made according to the content of each chapter to make it conform to the standardized financial report structure. After completing this series of adjustments, the final financial report (the second financial report) is formed.

[0089] Taking a manufacturing enterprise as an example, assume that there is a logical error in the enterprise's first financial report where revenue growth leads to a decline in net profit. Through systematic financial logic verification, the system identifies this problem and generates a correction task through data mapping and correction instructions, providing correction suggestions. Since the relationship between revenue and net profit is crucial for the enterprise's financial report, the priority of this correction task is high. The system processes it preferentially through a weighted correction task sequence and replaces the incorrect logical text with the statement that revenue growth and cost optimization jointly promote the increase in net profit. In the corrected report, the system further provides detailed correction basis for this adjustment, combining industry-standard cost control methods and financial data support from similar enterprises. At the same time, the system generates an analysis of the net profit fluctuations, pointing out that cost control is the key factor in the growth of net profit, and the text expression of the report is also adjusted to be standardized to conform to the standard financial report format. Through this series of data processing and corrections, the enterprise obtains a second financial report with a complete structure, self-consistent logic, and compliance with industry norms, which can provide more accurate financial analysis basis for decision-makers.

[0090] In a specific embodiment, the process of executing step S106 may specifically include the following steps:

[0091] Perform two-way cross-verification on the second financial report, reverse-verify the overall financial indicators with the segment data, and obtain the cross-verification result;

[0092] Input the second financial report into the knowledge base comparison mechanism, compare it with the historical financial report database, and identify the abnormal trend change points to obtain the set of trend abnormal points;

[0093] Conduct semantic framework analysis on the second financial report, detect the semantic mapping consistency between the financial narrative and the corresponding data, mark the areas where the narrative and data do not match, and obtain the semantic matching degree report;

[0094] Generate a problem space map based on the cross-verification result, the set of trend abnormal points, and the semantic matching degree report, locate the detected problem points in three-dimensional space and construct the problem evolution path to obtain the correction map;

[0095] Perform region-oriented updates on the second financial report according to the correction map, update the mutually influential financial indicators and their explanatory texts according to spatial relevance, and obtain the associated update report;

[0096] Perform intelligent integrity re-evaluation on the associated update report, establish the report content index structure, and obtain the target financial report.

[0097] Specifically, during the process of implementing financial report verification and optimization, the second financial report will be processed through a series of steps such as two-way cross-verification, knowledge base comparison, semantic framework analysis, problem space map generation, targeted update, and integrity re-evaluation. These steps ensure the accuracy and logical consistency of the financial report. Performing two-way cross-verification reversely verifies the overall financial indicators with the segment data. Specifically, the process of cross-verification is to compare and verify the data in each part of the financial report to ensure data consistency between the whole and the segments. For example, assume a company reports an overall revenue of 50 million yuan, and the revenue of a certain subsidiary in the segment data is 15 million yuan. When conducting cross-verification, the auditor needs to verify whether the sum of the overall revenue and the segment revenue is consistent. If there is an inconsistency, this issue will be marked as an anomaly, and the location and reason for its occurrence will be recorded. The second financial report will be input into the knowledge base comparison mechanism, compared with the historical financial report database, to identify abnormal trend change points and generate a set of trend anomaly points. In this process, the financial indicators involved in the report, such as revenue, gross profit, net profit, etc., will be extracted and compared with the same type of data in the historical financial reports. Through this comparison, abnormal fluctuations in certain financial data can be identified. For example, if historical data shows that the average gross profit margin of an industry is usually 40%, while the current report shows 10%, this constitutes an abnormal trend point. The identified set of trend anomaly points will provide a basis for subsequent financial analysis and report optimization, helping to analyze whether the changes in the company's financial situation are reasonable. Semantic framework analysis checks the semantic consistency between the financial narrative and the data. In financial reports, many data and their interpretations are linked through text descriptions. The task of semantic framework analysis is to ensure that these descriptions are consistent with the actual data. For example, if the report states that the company's net profit has increased by 20% compared to last year, but the net profit increase in the financial data is only 5%, then there is a mismatch between this description and the actual data. Through semantic framework analysis technology, the financial narrative and the data will be compared, and all inconsistent areas will be marked, and finally a semantic matching report will be generated to reveal potential language or data errors in the report.

[0098] Based on the cross-validation results, the set of trend anomaly points, and the semantic matching degree report, a problem space map will be generated to locate the problem points in the report and construct the problem evolution path. The problem space map is a three-dimensional coordinate system, and each problem point in it is located within the corresponding spatial region. This map can visually show the relationships between different problems in the financial report. For example, some problems may be caused by data errors, while others may be due to inconsistent financial narratives. Through this visualization method, financial reviewers can more clearly see the evolution process of the problems, that is, how a small error expands to logical inconsistencies throughout the report. These problem points and their evolution paths will help reviewers quickly find the key errors in the report and make corrections. The second financial report will be updated regionally and directionally according to the corrected map, and the financial indicators and their explanatory texts will be updated according to spatial relevance. Each problem point in the corrected map represents the area in the report that needs to be corrected, and based on the relevance between these areas, other related data and texts will be intelligently adjusted. For example, if an inconsistency is found in the description of the data item of revenue growth, the system will not only modify the description but also adjust the explanatory texts in other parts such as net profit or cost according to the relevant data of revenue growth. Through the directional update, the system ensures that each part of the report is adjusted in a timely and accurate manner, thus achieving the efficient correction of the financial report. After the associated update report is completed, an intelligent integrity re-evaluation will be performed, and an index structure of the report content will be established. The purpose of this process is to ensure that each part of the report has been reasonably corrected and that no new errors are introduced in the update of all data and texts. By establishing the index structure of the report content, each financial data item, each financial term, and its description can be indexed to ensure that the mutual relationships and logical consistency between data are not damaged during any modification of the report. After the integrity re-evaluation, the second financial report will become more accurate and standardized, forming the final target financial report.

[0099] The above described the method for generating a financial report based on artificial intelligence in the embodiments of the present application. Next, the system for generating a financial report based on artificial intelligence in the embodiments of the present application will be described. Please refer to Figure 2 , an embodiment of the system for generating a financial report based on artificial intelligence in the embodiments of the present application includes:

[0100] The acquisition module 201 is used to acquire and preprocess multi-source financial data to obtain target financial data;

[0101] The input module 202 is used to input the target financial data into a knowledge-based capital extraction algorithm for financial semantic understanding to obtain a financial data knowledge base;

[0102] A filling module 203, configured to perform intelligent report template matching and automatic filling of financial information according to the financial data knowledge base to obtain a first financial report;

[0103] A verification module 204, configured to perform financial logic verification on the first financial report through a causal relationship extraction model, identify the causal relationship between financial events and compare it with an expected model to obtain a financial logic verification report;

[0104] A generation module 205, configured to perform deep learning content generation and optimization on the first financial report based on the financial logic verification report to generate a second financial report;

[0105] An update module 206, configured to perform multi-level automatic review and intelligent update on the second financial report to obtain a target financial report.

[0106] Through the collaborative cooperation of the above-mentioned various components, through multi-source financial data collection and preprocessing, the problems of scattered data sources and inconsistent formats in the traditional financial report compilation process are solved, and the data processing efficiency is significantly improved; the knowledge-based capital extraction algorithm adopted breaks through the limitations of the traditional method's surface understanding of financial data, realizes the intelligent parsing of the deep semantics of financial data, accurately identifies the intellectual capital elements hidden in the financial data, and enables the financial report to comprehensively reflect the enterprise value; the intelligent report template matching and automatic filling technology overcomes the rigidity of traditional template use, dynamically selects and adjusts the optimal template according to the enterprise characteristics and financial data characteristics, and realizes the personalized customization of the report; and the application of the causal relationship extraction model is the core innovation point of the present invention. By integrating Chinese language characteristics and graph attention network technology, this model accurately identifies the causal logic relationship between financial events, effectively solves the problems of fuzzy causal relationships and logical contradictions in traditional financial reports, and makes the report content logically rigorous and well-organized; the content generation and optimization technology based on deep learning breaks through the limitations of single content and mechanical expression in traditional report generation methods, and intelligently generates professional and accurate explanatory texts according to the financial data characteristics and verification results, enhancing the readability and professional depth of the report; the multi-level automatic review and intelligent update technology constructs an all-round quality assurance system, and through multi-dimensional detections such as cross-verification, trend analysis and semantic matching, ensures the accuracy, integrity and standardization of the report, greatly reduces the report error rate, improves the generation efficiency of the financial report, and more importantly, ensures the quality and professional level of the report, providing an intelligent solution for the entire process of enterprise financial report compilation.

[0107] Above Figure 2 The financial report generation system based on artificial intelligence in the embodiments of the present invention is described in detail from the perspective of modular functional entities. Next, the financial report generation device based on artificial intelligence in the embodiments of the present invention is described in detail from the perspective of hardware processing.

[0108] Figure 3 It is a schematic structural diagram of a financial report generation device based on artificial intelligence provided by an embodiment of the present invention. The financial report generation device 300 based on artificial intelligence may vary greatly due to configuration or performance, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 for storing application programs 333 or data 332 (for example, one or more mass storage device terminals). Among them, the memory 320 and the storage media 330 may be transient storage or persistent storage. The program stored in the storage media 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations on the financial report generation device 300 based on artificial intelligence. Further, the processor 310 may be set to communicate with the storage media 330 and execute a series of instruction operations in the storage media 330 on the financial report generation device 300 to implement the steps of the above-mentioned financial report generation method based on artificial intelligence.

[0109] The financial report generation device 300 based on artificial intelligence may further include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input / output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, and so on. Those skilled in the art can understand that Figure 3 The shown structural diagram of the financial report generation device based on artificial intelligence does not limit the financial report generation device provided by the present invention, and may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0110] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are run on a computer, the computer is made to execute the steps of the financial report generation method based on artificial intelligence.

[0111] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described system, system, and unit can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0112] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing an artificial intelligence-based financial report generation device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0113] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of various embodiments of the present invention.

Claims

1. A method for generating financial reports based on artificial intelligence, characterized in that, The method includes: Collecting and preprocessing multi-source financial data to obtain target financial data; Inputting the target financial data into a knowledge-based capital extraction algorithm for financial semantic understanding to obtain a financial data knowledge base; Performing intelligent report template matching and automatic financial information filling based on the financial data knowledge base to obtain a first financial report; Verifying the financial logic of the first financial report through a causal relationship extraction model, identifying the causal relationships between financial events and comparing them with an expected model to obtain a financial logic verification report; Generating and optimizing the content of the first financial report through deep learning based on the financial logic verification report to generate a second financial report; Performing multi-level automatic review and intelligent update on the second financial report to obtain a target financial report.

2. The method for generating a financial report based on artificial intelligence according to claim 1, wherein The collecting and preprocessing of multi-source financial data to obtain target financial data includes: Regularly collecting the enterprise's core financial system, audit report library, historical financial statements, and external regulatory data to obtain an original financial data set; Performing format conversion processing on the original financial data set, converting the data in PDF statements, spreadsheets, and text documents into a standardized structure to obtain initial financial data; Checking the data integrity of the initial financial data, identifying missing items in key financial fields and filling them according to industry standards to obtain complete financial data; Performing consistency verification on the complete financial data, checking the articulation relationships between the balance sheet, income statement, and cash flow statement, and marking data items that do not conform to accounting standards to obtain verified financial data; Performing sensitive information processing on the verified financial data, grading and desensitizing customer identifiers and employee salary details to obtain desensitized financial data; Constructing a financial indicator association graph based on the desensitized financial data to obtain the target financial data.

3. The method for generating a financial report based on artificial intelligence according to claim 1, wherein The inputting of the target financial data into a knowledge-based capital extraction algorithm for financial semantic understanding to obtain a financial data knowledge base includes: Inputting the target financial data into the term recognition module in the knowledge-based capital extraction algorithm, and extracting financial keywords, amount values, and time information through a financial professional dictionary and named entity recognition technology to obtain a financial term set; Inputting the financial term set into the ontology mapping module in the knowledge-based capital extraction algorithm for processing, and mapping the terms to a standard financial concept system using a pre-set knowledge ontology in the financial field to obtain structured financial entities; Inputting the structured financial entities into the relationship extraction module in the knowledge-based capital extraction algorithm for financial semantic association analysis, and identifying the logical relationships between semantic subjects and objects through syntactic dependency analysis to obtain a financial matter relationship graph; Inputting the financial matter relationship graph into the event clustering module in the knowledge-based capital extraction algorithm for processing, and organizing related matters into event units according to financial semantic similarity and temporal correlation to obtain a financial event sequence; Inputting the financial event sequence into the capital classification module in the knowledge-based capital extraction algorithm for intellectual capital classification, and classifying and quantifying financial events according to a predefined capital category standard to obtain a classified financial knowledge structure; Input the classified financial knowledge structure into the weight calculation module in the knowledge-based capital extraction algorithm for processing. By calculating the importance weights of each knowledge node and constructing a multi-level index structure, the financial data knowledge base is obtained.

4. The method for generating a financial report based on artificial intelligence according to claim 1, wherein Performing intelligent report template matching and automatic financial information filling based on the financial data knowledge base to obtain the first financial report, including: Extracting enterprise asset scale, liability structure, and profit indicator data from the financial data knowledge base to obtain enterprise financial characteristic data; Calculating the template matching degree in the template library based on the enterprise financial characteristic data, and selecting the template with the highest matching degree to obtain the preferred report template; Conducting component evaluation on the preferred report template, selecting table components suitable for enterprise data and combining them to obtain a customized report structure; Establishing a mapping relationship between the fields of the customized report structure and the data nodes in the financial data knowledge base to obtain field filling rules; Filling the data in the financial data knowledge base into the customized report structure according to the field filling rules, calculating financial indicators and filling the results to obtain a complete report; Performing format standardization processing on the complete report, adjusting numerical formats, table styles, and marking areas that need to be supplemented and explained to obtain the first financial report.

5. The method for generating a financial report based on artificial intelligence according to claim 1, wherein Verifying the financial logic of the first financial report through a causal relationship extraction model, identifying the causal relationships between financial events and comparing them with the expected model to obtain a financial logic verification report, including: Decomposing the first financial report into multiple financial event units, and extracting key financial indicator changes and business activities through word segmentation, syntactic parsing, and named entity recognition to obtain a financial event set; Constructing a financial event graph for the financial event set, taking each financial event as a node, and initializing the potential connection relationships between nodes to obtain an initial financial event network; Inputting the initial financial event network into the Chinese dictionary enhancement layer in the causal relationship extraction model, and enhancing the semantics of event descriptions through a financial professional dictionary to obtain a semantically enhanced event representation; Performing financial event dependency analysis on the semantically enhanced event representation through the graph attention network layer in the causal relationship extraction model, calculating the multi-head attention weights between nodes, and identifying the causal association strength between events to obtain a weighted causal relationship graph; Comparing the weighted causal relationship graph with the preset causal knowledge base in the financial field, detecting abnormal causal links and logical contradictions, and marking causal relationships that do not conform to financial laws to obtain financial logic anomaly marks; Generating a financial logic verification report based on the financial logic anomaly marks, including causal chains that pass the verification and abnormal causal relationships that need to be corrected, and providing correction suggestions for each anomaly to obtain the financial logic verification report.

6. The method for generating a financial report based on artificial intelligence according to claim 1, wherein Generating and optimizing the content of the first financial report through deep learning based on the financial logic verification report to generate the second financial report, including: Mapping the data of the financial logic verification report and the first financial report, extracting the position marks and corresponding correction suggestions for each area that needs to be corrected to obtain a correction instruction data set; Sort the corrected instruction data set according to the anomaly degree quantization value, calculate the priority score of each correction task, and obtain a weighted correction task sequence; Based on the weighted correction task sequence, perform content reconstruction on the first financial report, replace the text in the unreasonable financial logic area, and supplement the explanation of the correction basis to obtain a draft logical correction report; Generate key indicator explanations for the draft logical correction report, analyze the reasons for abnormal fluctuations in financial indicators, and add professional background analysis text to obtain an enhanced report on indicator explanations; Perform text expression conversion on the enhanced report on indicator explanations to obtain a text normalization report; Adjust the chapter organization of the text normalization report to obtain the second financial report.

7. The method for generating a financial report based on artificial intelligence according to claim 1, wherein Perform multi-level automatic review and intelligent update on the second financial report to obtain the target financial report, including: Perform two-way cross-review on the second financial report, reverse verify the overall financial indicators with the segment data, and obtain the cross-verification result; Input the second financial report into the knowledge base comparison mechanism, compare it with the historical financial report database, and identify the abnormal trend change points to obtain a set of trend anomaly points; Perform semantic framework analysis on the second financial report, detect the semantic mapping consistency between the financial narrative and the corresponding data, and mark the areas where the narrative and data do not match to obtain a semantic matching degree report; Generate a problem space map based on the cross-verification result, the set of trend anomaly points, and the semantic matching degree report, locate the detected problem points in three-dimensional space and construct a problem evolution path to obtain a correction map; Perform region-oriented update on the second financial report according to the correction map, update the mutually influential financial indicators and their explanatory texts according to the spatial relevance to obtain an associated update report; Perform intelligent integrity re-evaluation on the associated update report, establish an index structure for the report content to obtain the target financial report.

8. A financial report generation system based on artificial intelligence, characterized in that, For implementing the artificial intelligence-based financial report generation method according to any one of claims 1-7, the artificial intelligence-based financial report generation system includes: A collection module for collecting and preprocessing multi-source financial data to obtain target financial data; An input module for inputting the target financial data into a knowledge-based capital extraction algorithm for financial semantic understanding to obtain a financial data knowledge base; A filling module for performing intelligent report template matching and automatic filling of financial information according to the financial data knowledge base to obtain a first financial report; A verification module for verifying the financial logic of the first financial report through a causal relationship extraction model, identifying the causal relationship between financial events and comparing it with the expected model to obtain a financial logic verification report; A generation module for performing deep learning content generation and optimization on the first financial report based on the financial logic verification report to generate a second financial report; An update module for performing multi-level automatic review and intelligent update on the second financial report to obtain the target financial report.

9. A financial report generation device based on artificial intelligence, characterized in that, It includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the computer program, it implements the artificial intelligence-based financial report generation method described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is run by the processor, it causes the processor to execute the artificial intelligence-based financial report generation method described in any one of claims 1 to 7.

Images