AI-based automated enterprise financial analysis system and method

Abstract

The present invention relates to the technical field of enterprise financial analysis, and specifically discloses an automated enterprise financial analysis system and method based on AI, wherein the system comprises: a data acquisition module, used for acquiring effective financial data of an enterprise to be analyzed based on AI; a knowledge graph construction module, used for constructing a financial concept knowledge graph; a data parsing module, used for parsing the effective financial data of the enterprise to be analyzed based on the financial concept knowledge graph, and obtaining a data graph based on the financial analysis of the enterprise to be analyzed; a decision suggestion module, used for obtaining decision suggestions corresponding to the financial analysis decision purpose based on the financial analysis decision purpose and the data graph based on the financial analysis of the enterprise to be analyzed; and used for performing fully automated in-depth analysis and intelligent decision support on the enterprise financial data based on the constructed financial concept knowledge graph, thereby improving the data parsing efficiency and the reliability of decision suggestions in the financial analysis process.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise financial analysis, and in particular to an AI-based automated enterprise financial analysis system and method. Background Art

[0002] With the deepening of digital transformation, enterprises have a growing demand for efficient and accurate financial data analysis. At present, there are some financial management software and service providers on the market, which have realized the automated processing and analysis of financial data to a certain extent. However, these solutions still have some shortcomings.

[0003] However, facing massive amounts of financial data, especially unstructured data (such as financial reports in PDF format), traditional financial management systems find it difficult to effectively standardize and convert financial data. Data cleaning and conversion efficiency is low, and data quality is difficult to guarantee, resulting in uneven data quality and difficulty in meeting the requirements of high-quality data analysis. Existing financial analysis systems usually use fixed analysis models, lacking flexibility and scalability. In addition, these systems can often only provide simple indicator calculations, but it is difficult to achieve in-depth business insights. For complex analysis scenarios, such as asset quality assessment and profit quality analysis, existing tools are often unable to cope with them. Due to the fixed analysis model, it is difficult to adapt to the characteristics of different industries and enterprises; the in-depth analysis capabilities are limited and cannot provide comprehensive business insights. In summary, although the existing automated financial management tools have improved work efficiency to a certain extent, there is still room for improvement in terms of the quality of data collection, the flexibility and depth of data analysis, etc. Therefore, there is an urgent need for a new automated financial management solution that can overcome the above defects.

[0004] Therefore, the present invention proposes an AI-based automated enterprise financial analysis system and method. Summary of the invention

[0005] The present invention provides an AI-based automated enterprise financial analysis system and method, which is used to obtain effective financial data of the enterprise to be analyzed based on AI, effectively standardize and convert the financial data, and ensure the quality and effectiveness of the financial data; and introduces a knowledge graph to construct a financial concept knowledge graph, and performs fully automated in-depth analysis and intelligent decision support on the enterprise financial data based on the constructed financial concept knowledge graph. Compared with traditional financial management tools, it does not rely solely on fixed analysis models, improves the flexibility and scalability of data parsing in the financial analysis process, and can achieve in-depth business insights, thereby providing timely and accurate financial analysis conclusions and decision support in a fully automatic mode, thereby improving the efficiency of enterprise financial analysis, the quality of data collection, and the flexibility and depth of data analysis.

[0006] The present invention provides an AI-based automated enterprise financial analysis system, comprising:

[0007] Data acquisition module, used to obtain valid financial data of the enterprise to be analyzed based on AI;

[0008] Knowledge graph construction module, used to build financial concept knowledge graph;

[0009] The data analysis module is used to analyze the effective financial data of the enterprise to be analyzed based on the financial concept knowledge map, and obtain the data map based on the financial analysis of the enterprise to be analyzed;

[0010] The decision-making suggestion module is used to obtain decision suggestions corresponding to the financial analysis decision purpose based on the financial analysis decision purpose and the financial analysis basis data map of the enterprise to be analyzed.

[0011] Preferably, the data acquisition module includes:

[0012] The financial data collection submodule is used to collect the financial statements, transaction records, and market data of the enterprise to be analyzed and summarize them as the financial data of the enterprise to be analyzed;

[0013] The data cleaning and verification submodule is used to clean and verify the financial data of the company to be analyzed based on AI to obtain valid financial data of the company to be analyzed.

[0014] Preferably, the knowledge graph construction module includes:

[0015] The indicator naming submodule is used to treat asset quality, income quality, profit quality, cash flow quality and labor efficiency quality as first-level indicators, and determine all measurement dimensions of each first-level indicator as all second-level indicators of each first-level indicator, and all specific measurement indicators of each measurement dimension as all third-level indicators of each second-level indicator;

[0016] The tree structure building submodule is used to treat all the first-level indicators as all the root nodes of the tree structure, and at the same time, treat all the second-level indicators corresponding to each first-level indicator as all the child nodes of the corresponding root node in the corresponding tree structure, and at the same time, treat all the third-level indicators of each second-level indicator as all the leaf nodes of the corresponding child nodes in the corresponding tree structure, so as to obtain at least one financial concept tree structure map;

[0017] The association line supplement submodule is used to set an additional calculation association line between the corresponding two third-level indicators to obtain a financial concept knowledge graph when it is determined that there is at least one third-level indicator in all financial concept tree structure graphs that is an intermediate calculation quantity of another third-level indicator.

[0018] Preferably, the data parsing module includes:

[0019] The final indicator screening submodule is used to treat all the third-level indicators in the financial concept knowledge graph that do not have another third-level indicator as the corresponding intermediate calculation amount as the real final-level indicators;

[0020] The original parameter class selection submodule is used to obtain all the original parameter classes of each real final-level indicator in the enterprise to be analyzed based on the final basis calculation method of each real final-level indicator;

[0021] The graph supplement submodule is used to compile all original parameter classes of each real final-level indicator into the financial concept knowledge graph as subnodes of the corresponding node in the financial concept knowledge graph, so as to obtain a refined financial concept knowledge graph of the enterprise to be analyzed;

[0022] The graph simplification submodule is used to merge the same original parameter classes in the refined financial concept knowledge graph of the enterprise to be analyzed to obtain the exclusive financial concept knowledge graph of the enterprise to be analyzed;

[0023] The data analysis and deduction submodule is used to analyze and deduce the effective financial data of the enterprise to be analyzed based on the exclusive financial concept knowledge graph of the enterprise to be analyzed, and obtain the data graph based on the financial analysis of the enterprise to be analyzed.

[0024] Preferably, the original parameter class selection submodule includes:

[0025] An application feasibility evaluation unit is used to determine all possible calculation methods for each third-level indicator and evaluate the application feasibility of each possible calculation method for each third-level indicator for the enterprise to be analyzed;

[0026] The inspection object distinguishing unit is used to summarize the real final-level indicator with another third-level indicator as the corresponding intermediate calculation amount and the third-level indicator corresponding to the corresponding intermediate calculation amount as the first inspection object, and to take each third-level indicator among all the third-level indicators except all the third-level indicators included in the first inspection object as the second inspection object;

[0027] A first calculation method screening unit is used to screen out the final basis calculation methods corresponding to the two third-level indicators in each first investigation object from all the basis calculation methods of the two third-level indicators in each first investigation object based on the feasibility of each basis calculation method of each third-level indicator for the enterprise to be analyzed;

[0028] A second calculation method screening unit is used to use the calculation method corresponding to the maximum application feasibility among all the calculation methods that can be used for each second investigation object as the final calculation method for the third-level indicator corresponding to each second investigation object;

[0029] The original parameter class selection unit is used to regard all the original parameter classes required in the final basis calculation method of each real final-level indicator as all the original parameter classes corresponding to the real final-level indicator.

[0030] Preferably, the application feasibility assessment unit includes:

[0031] The calculation method determination subunit is used to determine all the available calculation methods for each third-level indicator;

[0032] The similar enterprise screening subunit is used to screen out all reference enterprises whose scale similarity with the enterprise to be analyzed is not less than the similarity threshold in the reference enterprise financial analysis database as all enterprises with similar scales to the enterprise to be analyzed;

[0033] The application feasibility calculation subunit is used to take the ratio of the total application times of each available calculation method of each third-level indicator in all enterprises of similar size to the enterprise to be analyzed and the total application times of all available calculation methods of the corresponding third-level indicator in all enterprises of similar size to the enterprise to be analyzed, as the application feasibility degree of each available calculation method of each third-level indicator for the enterprise to be analyzed.

[0034] Preferably, the first calculation method screening unit includes:

[0035] The application feasibility calculation subunit is used to determine all interrelated calculation method groups among all available calculation methods of the two third-level indicators in the first investigation object, and calculate the actual application feasibility of each interrelated calculation method group based on the application feasibility of each available calculation method of the third-level indicators for the enterprise to be analyzed;

[0036] The final calculation method screening subunit is used to use the two available calculation methods contained in the interrelated calculation method group with the greatest practical application feasibility among all interrelated calculation method groups of each first investigation object as the final basis calculation methods for the two third-level indicators in each first investigation object.

[0037] Preferably, the data analysis and derivation submodule includes:

[0038] The original parameter class group determination unit is used to group all the original parameter classes of each real final-level indicator in the exclusive financial concept knowledge graph of the enterprise to be analyzed to obtain multiple original parameter class groups;

[0039] A tree structure reverse derivation unit is used to reversely deduce all associations in the financial concept knowledge graph with each original parameter class group in the exclusive financial concept knowledge graph of the enterprise to be analyzed as a starting point, and obtain the data derivation logic of each original parameter class group, wherein the data derivation logic is a single-threaded data derivation logic or a multi-threaded data derivation logic;

[0040] The data analysis and derivation unit is used to analyze and derive the effective financial data of the enterprise to be analyzed based on the data derivation logic of all original parameter class groups, and obtain the analysis and derivation results of each first-level indicator, second-level indicator, and third-level indicator in the exclusive financial concept knowledge graph;

[0041] The analytical derivation results of all the first-level indicators, second-level indicators, and third-level indicators in the exclusive financial concept knowledge graph are bound to the corresponding nodes in the exclusive financial concept knowledge graph to obtain the data graph based on the financial analysis of the enterprise to be analyzed.

[0042] Preferably, the decision suggestion module includes:

[0043] The derivation path fitting submodule is used to fit the decision derivation path in the financial analysis basis data map of the enterprise to be analyzed based on the decision logic corresponding to the financial analysis decision purpose;

[0044] The data re-derivation submodule is used to re-derive the node data in the data map of the financial analysis basis of the analyzed enterprise based on the decision derivation path, and obtain the final decision basis for the financial analysis decision purpose;

[0045] The decision suggestion acquisition submodule is used to obtain the decision suggestion corresponding to the financial analysis decision purpose based on the final decision basis of the financial analysis decision purpose.

[0046] The present invention provides an AI-based automated enterprise financial analysis method, which is applied to any of the above-mentioned AI-based automated enterprise financial analysis systems, including:

[0047] S1: Obtain valid financial data of the company to be analyzed based on AI;

[0048] S2: Constructing a knowledge graph of financial concepts;

[0049] S3: Analyze the effective financial data of the enterprise to be analyzed based on the financial concept knowledge graph to obtain the data graph based on the financial analysis of the enterprise to be analyzed;

[0050] S4: Based on the decision-making purpose of financial analysis and the data map of the financial analysis basis of the enterprise to be analyzed, obtain decision recommendations corresponding to the decision-making purpose of financial analysis.

[0051] The beneficial effects of the present invention compared with the prior art are as follows: based on AI, effective financial data of the enterprise to be analyzed is obtained, and the financial data is effectively standardized and converted, thereby ensuring the quality and effectiveness of the financial data; and a knowledge graph is introduced to construct a financial concept knowledge graph, and based on the constructed financial concept knowledge graph, the enterprise financial data is fully automated for in-depth analysis and intelligent decision support. Compared with traditional financial management tools, it does not rely solely on fixed analysis models, improves the flexibility and scalability of data analysis in the financial analysis process, and can achieve in-depth business insights, thereby providing timely and accurate financial analysis conclusions and decision support in a fully automatic mode, thereby improving the efficiency of enterprise financial analysis, the quality of data collection, and the flexibility and depth of data analysis.

[0052] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in this application document.

[0053] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0055] Figure 1 Schematic diagram of internal functional modules of an AI-based automated enterprise financial analysis system in an embodiment of the present invention;

[0056] Figure 2 Schematic diagram of internal functional submodules of a data acquisition module in an embodiment of the present invention;

[0057] Figure 3 A schematic diagram of the internal functional submodules of the knowledge graph construction module in an embodiment of the present invention;

[0058] Figure 4 Schematic diagram of internal functional submodules of a data analysis module in an embodiment of the present invention;

[0059] Figure 5 A schematic diagram of internal functional submodules of a decision suggestion module in an embodiment of the present invention;

[0060] Figure 6 This is a flow chart of an AI-based automated enterprise financial analysis method in an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0062] Embodiment 1:

[0063] The present invention provides an automated enterprise financial analysis system based on AI, referring to Figure 1 ,include:

[0064] Data acquisition module, used to obtain valid financial data of the enterprise to be analyzed based on AI;

[0065] Knowledge graph construction module, used to build financial concept knowledge graph;

[0066] The data analysis module is used to analyze the effective financial data of the enterprise to be analyzed based on the financial concept knowledge map, and obtain the data map based on the financial analysis of the enterprise to be analyzed;

[0067] The decision-making suggestion module is used to obtain decision suggestions corresponding to the financial analysis decision purpose based on the financial analysis decision purpose and the financial analysis basis data map of the enterprise to be analyzed.

[0068] In this embodiment, the enterprise to be analyzed is an enterprise that needs to use this embodiment to perform automated enterprise financial analysis.

[0069] In this embodiment, the valid financial data is the cleansed and verified financial data of the enterprise to be analyzed.

[0070] In this embodiment, the financial concept knowledge graph includes a knowledge graph of financial concepts (i.e., primary indicators, secondary indicators, tertiary indicators, etc.) and the relationships between them.

[0071] In this embodiment, the financial analysis is based on the financial concept knowledge graph, and each node contained in the data graph is bound to the knowledge graph of the financial performance data of the enterprise to be analyzed on the corresponding indicator.

[0072] In this embodiment, the purposes of financial analysis and decision-making include risk warning, investment advice, cost control, etc.

[0073] In this embodiment, the decision suggestion is an improvement suggestion that can achieve the purpose of financial analysis decision making.

[0074] The beneficial effects of the above technologies are: based on AI, effective financial data of the company to be analyzed is obtained, and the financial data is effectively standardized and converted, ensuring the quality and effectiveness of the financial data; and a knowledge graph is introduced to construct a financial concept knowledge graph, and based on the constructed financial concept knowledge graph, the company's financial data is fully automated for in-depth analysis and intelligent decision support. Compared with traditional financial management tools, it does not rely solely on fixed analysis models, but improves the flexibility and scalability of data analysis in the financial analysis process, and can achieve in-depth business insights, thereby providing timely and accurate financial analysis conclusions and decision support in a fully automatic mode, improving the efficiency of corporate financial analysis, the quality of data collection, and the flexibility and depth of data analysis.

[0075] Example 2

[0076] Based on Example 1, the data acquisition module, refer to Figure 2 ,include:

[0077] The financial data collection submodule is used to collect the financial statements, transaction records, and market data of the enterprise to be analyzed and summarize them as the financial data of the enterprise to be analyzed;

[0078] The data cleaning and verification submodule is used to clean and verify the financial data of the company to be analyzed based on AI to obtain valid financial data of the company to be analyzed.

[0079] In this embodiment, the market data mainly includes information such as price, demand, supply, consumption habits, etc. of various commodities or services in the market.

[0080] In this embodiment, the financial data of the enterprise to be analyzed is cleaned and verified based on AI to obtain valid financial data of the enterprise to be analyzed, including:

[0081] The financial data of the enterprise to be analyzed that is directly collected is input into the pre-trained AI data cleaning and verification model to obtain the valid financial data of the enterprise to be analyzed, wherein the pre-trained AI data cleaning and verification model is, for example:

[0082] Model structure: linear regression + LSTM;

[0083] Input variables: Financial data such as historical sales, advertising expenditures, and website visits;

[0084] Output variables: valid historical sales, advertising expenditures, website visits and other financial data that have been manually cleaned and verified;

[0085] The training process includes:

[0086] Data collection: Collect sales data over a period of time, as well as other possible predictive factors such as advertising expenditures, website visits, etc.

[0087] Data preprocessing: Preprocess the data, including removing null values, outliers, missing values, etc. For time series data, you can use the resample function in pandas to convert it to StationaryTimeSeries format;

[0088] Divide the dataset: Divide the dataset into training and test sets based on past sales data;

[0089] Model training: Use the training set data to train the LSTM model and set hyperparameters such as learning_rate, window_size, etc.

[0090] Model evaluation: Use the test set data to evaluate the model and calculate the model's mean square error (MSE), mean absolute error (MAE), root mean square error (RMSE) and other indicators;

[0091] Model optimization: Based on the results of model evaluation, optimize the model, such as adjusting hyperparameters, changing model structure, etc.

[0092] The beneficial effects of the above technology are: clarifying the specific data items of the financial data of the enterprise to be analyzed, and introducing a pre-trained AI data cleaning and verification model to achieve effective cleaning and verification of the financial data of the enterprise to be analyzed.

[0093] Example 3

[0094] Based on Example 1, the knowledge graph construction module is referenced Figure 3 ,include:

[0095] The indicator naming submodule is used to treat asset quality, income quality, profit quality, cash flow quality and labor efficiency quality as first-level indicators, and determine all measurement dimensions of each first-level indicator as all second-level indicators of each first-level indicator, and all specific measurement indicators of each measurement dimension as all third-level indicators of each second-level indicator;

[0096] The tree structure building submodule is used to treat all the first-level indicators as all the root nodes of the tree structure, and at the same time, treat all the second-level indicators corresponding to each first-level indicator as all the child nodes of the corresponding root node in the corresponding tree structure, and at the same time, treat all the third-level indicators of each second-level indicator as all the leaf nodes of the corresponding child nodes in the corresponding tree structure, so as to obtain at least one financial concept tree structure map;

[0097] The association line supplement submodule is used to set an additional calculation association line between the corresponding two third-level indicators to obtain a financial concept knowledge graph when it is determined that there is at least one third-level indicator in all financial concept tree structure graphs that is an intermediate calculation quantity of another third-level indicator.

[0098] In this embodiment, all measurement dimensions of the primary indicator are measurement dimensions used to measure the performance of the financial status of the enterprise to be analyzed in terms of the primary indicator. For example, when the primary indicator is asset quality, all corresponding measurement dimensions include asset liquidity, asset utilization efficiency, rationality of asset structure, etc.

[0099] For example, when the first-level indicator is income quality, all its corresponding measurement dimensions include: income growth, income stability, income sustainability, etc.

[0100] In this embodiment, all specific measurement indicators of each measurement dimension are calculation indicators that can measure the specific expression status of the enterprise to be analyzed in the first-level indicator from the corresponding measurement dimension, for example:

[0101] Specific indicators for measuring asset liquidity include: asset current ratio, asset quick ratio, asset cash ratio, etc.;

[0102] For example, specific indicators for measuring asset utilization efficiency include inventory turnover rate, accounts receivable turnover days, etc.

[0103] In this embodiment, the additional calculation association line is a node connection line contained in the financial concept knowledge graph that represents the association relationship between the three-level indicators (the association relationship marks the calculation method of using one indicator to calculate another indicator).

[0104] The beneficial effects of the above technology are: clarifying the specific meanings of all first-level indicators, second-level indicators, and third-level indicators and the correlation between them, and then building a financial concept knowledge graph.

[0105] Example 4

[0106] Based on Example 1, the data analysis module refers to Figure 4 ,include:

[0107] The final indicator screening submodule is used to treat all the third-level indicators in the financial concept knowledge graph that do not have another third-level indicator as the corresponding intermediate calculation amount as the real final-level indicators;

[0108] The original parameter class selection submodule is used to obtain all the original parameter classes of each real final-level indicator in the enterprise to be analyzed based on the final basis calculation method of each real final-level indicator;

[0109] The graph supplement submodule is used to compile all original parameter classes of each real final-level indicator into the financial concept knowledge graph as subnodes of the corresponding node in the financial concept knowledge graph, so as to obtain a refined financial concept knowledge graph of the enterprise to be analyzed;

[0110] The graph simplification submodule is used to merge the same original parameter classes in the refined financial concept knowledge graph of the enterprise to be analyzed to obtain the exclusive financial concept knowledge graph of the enterprise to be analyzed;

[0111] The data analysis and deduction submodule is used to analyze and deduce the effective financial data of the enterprise to be analyzed based on the exclusive financial concept knowledge graph of the enterprise to be analyzed, and obtain the data graph based on the financial analysis of the enterprise to be analyzed.

[0112] In this embodiment, the final calculation method of the real final-level indicator is the calculation method used when calculating the specific value of the real final-level indicator using financial data.

[0113] In this embodiment, all original parameter classes of the real final-level indicators are all parameter classes contained in the financial data and used to calculate the specific values ​​of the real final-level indicators, such as the total sales of the latest year, the total profit of the latest year, etc.

[0114] The beneficial effects of the above technology are: the real final-level indicators are screened out from all the third-level indicators in the financial concept knowledge graph, and all the original parameter classes of each real final-level indicator are combined to generate an exclusive financial concept knowledge graph for the enterprise to be analyzed, which provides a basis for analyzing the effective financial data of the enterprise to be analyzed, and based on the analytical deduction process of the effective financial data of the enterprise to be analyzed, the data graph based on which the financial analysis of the enterprise to be analyzed is obtained.

[0115] Example 5

[0116] Based on Example 4, the original parameter class selection submodule includes:

[0117] An application feasibility evaluation unit is used to determine all possible calculation methods for each third-level indicator and evaluate the application feasibility of each possible calculation method for each third-level indicator for the enterprise to be analyzed;

[0118] The inspection object distinguishing unit is used to summarize the real final-level indicator with another third-level indicator as the corresponding intermediate calculation amount and the third-level indicator corresponding to the corresponding intermediate calculation amount as the first inspection object, and to take each third-level indicator among all the third-level indicators except all the third-level indicators included in the first inspection object as the second inspection object;

[0119] A first calculation method screening unit is used to screen out the final basis calculation methods corresponding to the two third-level indicators in each first investigation object from all the basis calculation methods of the two third-level indicators in each first investigation object based on the feasibility of each basis calculation method of each third-level indicator for the enterprise to be analyzed;

[0120] A second calculation method screening unit is used to use the calculation method corresponding to the maximum application feasibility among all the calculation methods that can be used for each second investigation object as the final calculation method for the third-level indicator corresponding to each second investigation object;

[0121] The original parameter class selection unit is used to regard all the original parameter classes required in the final basis calculation method of each real final-level indicator as all the original parameter classes corresponding to the real final-level indicator.

[0122] In this embodiment, all available calculation methods for each third-level indicator are obtained through web crawling technology, for example:

[0123] The inventory turnover rate can be calculated by the method of calculating the inventory turnover times or the inventory turnover days in the prior art, including:

[0124] Inventory turnover rate is an indicator that measures the speed of inventory turnover in a company over a certain period of time. It reflects the company's sales ability and inventory management efficiency. Inventory turnover rate can be calculated in the following two ways:

[0125] 1. Inventory turnover rate (inventory turnover rate):

[0126] Inventory turnover refers to the average number of times an enterprise's inventory is sold and replaced in a certain period of time. The calculation formula is:

[0127] Inventory turnover = cost of goods sold (COGS) / average inventory;

[0128] Among them, cost of goods sold (COGS) refers to the cost of goods sold in a certain period, and average inventory refers to the average value of inventory in a certain period (usually one year), which can be calculated by the average of the beginning inventory and the ending inventory, that is:

[0129] Average inventory = (beginning inventory + ending inventory) / 2;

[0130] 2. Inventory turnover days (inventory turnover rate):

[0131] Inventory turnover days refers to the number of days required for the entire cycle from obtaining inventory to realizing sales and collecting cash. The calculation formula is:

[0132] Inventory turnover days = 365 (or 360) / inventory turnover times;

[0133] Or directly using average inventory and cost of goods sold:

[0134] Inventory turnover days = average inventory × 365 (or 360) / cost of goods sold;

[0135] In this embodiment, the calculation method of the third-level index is all the methods recorded in the prior art that can calculate the third-level index.

[0136] In this embodiment, the feasibility of each calculation method of each third-level indicator for the enterprise to be analyzed represents the feasibility of each calculation method of each third-level indicator when analyzing the financial data of the enterprise to be analyzed.

[0137] In this embodiment, the first investigation object includes the third-level indicator A and the real final-level indicator B of the intermediate calculation amount corresponding to the third-level indicator A.

[0138] The beneficial effects of the above technology are: by classifying all the third-level indicators into the first investigation objects and the second investigation objects, the calculation methods that can be used for the third-level indicators that are mutually correlated and the third-level indicators that are not correlated among the three-level indicators are screened according to different situations, and combined with the assessed feasibility of each calculation method for each third-level indicator for the enterprise to be analyzed, the final calculation method with the highest degree of comprehensive feasibility for each third-level indicator in the financial analysis process of the enterprise to be analyzed is screened out, and based on this, all the original parameter classes of each real final-level indicator are determined.

[0139] Example 6

[0140] On the basis of Embodiment 5, a feasibility evaluation unit is applied, including:

[0141] The calculation method determination subunit is used to determine all the available calculation methods for each third-level indicator;

[0142] The similar enterprise screening subunit is used to screen out all reference enterprises whose scale similarity with the enterprise to be analyzed is not less than the similarity threshold in the reference enterprise financial analysis database as all enterprises with similar scales to the enterprise to be analyzed;

[0143] The application feasibility calculation subunit is used to take the ratio of the total application times of each available calculation method of each third-level indicator in all enterprises of similar size to the enterprise to be analyzed and the total application times of all available calculation methods of the corresponding third-level indicator in all enterprises of similar size to the enterprise to be analyzed, as the application feasibility degree of each available calculation method of each third-level indicator for the enterprise to be analyzed.

[0144] In this embodiment, the calculation method of the similarity between the scale of the reference enterprise and the enterprise to be analyzed is:

[0145] To calculate the similarity between the sizes of two companies, we first need to determine the standard used to measure the size of the companies. Common standards include operating income, total assets, number of employees, etc. Depending on the standard chosen, different methods can be used to quantify the similarity between the two companies. Here are some common methods:

[0146] 1. Direct comparison method:

[0147] The simplest way is to directly compare the numerical differences between two companies on a certain indicator, such as operating income, total assets, etc. If the values ​​are close, then it can be considered that they have a high degree of similarity in this aspect.

[0148] Calculation formula:

[0149] Similarity = (the index value of enterprise A - the index value of enterprise B) / (the index value of enterprise A + the index value of enterprise B) after taking the absolute value;

[0150] This formula calculates the ratio of the difference between the two companies' index values ​​to the total. The smaller the value, the smaller the difference between the two companies in this index, that is, the closer their scales are.

[0151] 2. Proportional method:

[0152] Ratios can be used to assess the similarity of size between two businesses. For example, if the ratio of operating income between two businesses is very close to 1:1, then they can be considered to have a high degree of similarity in this respect.

[0153] In this embodiment, the reference enterprise financial analysis database includes a database of financial analysis examples in which financial data of a large number of enterprises of different sizes are manually parsed and calculated to obtain different analysis results for each third-level indicator.

[0154] In this embodiment, the similarity threshold is a preset value that the size similarity between the enterprise to be analyzed and the enterprise to be analyzed cannot be less than.

[0155] In this embodiment, the reference enterprise is an enterprise that has been subjected to financial analysis and is included in the reference enterprise financial analysis database.

[0156] The beneficial effect of the above technology is: by screening out enterprises of similar size to the enterprise to be analyzed in the reference enterprise financial analysis database, and based on the total number of applications of each available calculation method of each third-level indicator in all enterprises of similar size to the enterprise to be analyzed, the feasibility of application of each available calculation method of each third-level indicator to the enterprise to be analyzed is accurately calculated.

[0157] Embodiment 7:

[0158] Based on Example 5, the first calculation method screening unit includes:

[0159] The application feasibility calculation subunit is used to determine all interrelated calculation method groups among all available calculation methods of the two third-level indicators in the first investigation object, and calculate the actual application feasibility of each interrelated calculation method group based on the application feasibility of each available calculation method of the third-level indicators for the enterprise to be analyzed;

[0160] The final calculation method screening subunit is used to use the two available calculation methods contained in the interrelated calculation method group with the greatest practical application feasibility among all interrelated calculation method groups of each first investigation object as the final basis calculation methods for the two third-level indicators in each first investigation object.

[0161] In this embodiment, the mutual correlation calculation method group includes two mutual correlation calculation methods of the third-level indicators in the first investigation object, for example:

[0162] "Average inventory" is the intermediate quantity of "inventory turnover times". "Average inventory" and "inventory turnover times" belong to the first inspection object;

[0163] Among them, the calculation methods of "inventory turnover days" are:

[0164] Inventory turnover = cost of goods sold (COGS) / average inventory

[0165] The calculation methods of “average inventory” are:

[0166] Average inventory = (beginning inventory + ending inventory) / 2;

[0167] When the "beginning inventory" and "ending inventory" in the calculation method of "average inventory" are for quarters (i.e., corresponding to "beginning inventory at quarter" and "end inventory at quarter"), the calculation method of "inventory turnover days" corresponding to the "cost of goods sold (COGS)" in the above "inventory turnover days" calculation formula is for quarters (meaning: cost of goods sold (COGS) is the product of the sales quantity in the corresponding quarter and the unit cost of the corresponding batch). The calculation method of "average inventory" corresponding to the "beginning inventory" and "ending inventory" in the above "average inventory" calculation formula when they are for quarters are two interrelated calculation methods.

[0168] In this embodiment, based on the feasibility of each calculation method of the three-level indicators for the enterprise to be analyzed, the actual feasibility of each group of interrelated calculation methods is calculated, including:

[0169] The average of the application feasibility of the two calculation methods of each interrelated calculation method group to the enterprise to be analyzed is taken as the actual application feasibility of each interrelated calculation method group.

[0170] In this embodiment, the actual application feasibility of the interrelated calculation method group refers to the application feasibility of two calculation methods included in the interrelated calculation method group when they can be simultaneously applied in the financial analysis process of the enterprise to be analyzed.

[0171] The beneficial effects of the above technology are: accurately calculating the practical application feasibility of each group of interrelated calculation methods, and determining the final basis calculation method for the two third-level indicators in each first investigation object.

[0172] Example 8

[0173] Based on Example 4, the data analysis and derivation submodule includes:

[0174] The original parameter class group determination unit is used to group all the original parameter classes of each real final-level indicator in the exclusive financial concept knowledge graph of the enterprise to be analyzed to obtain multiple original parameter class groups;

[0175] A tree structure reverse derivation unit is used to reversely deduce all associations in the financial concept knowledge graph with each original parameter class group in the exclusive financial concept knowledge graph of the enterprise to be analyzed as a starting point, and obtain the data derivation logic of each original parameter class group, wherein the data derivation logic is a single-threaded data derivation logic or a multi-threaded data derivation logic;

[0176] The data analysis and derivation unit is used to analyze and derive the effective financial data of the enterprise to be analyzed based on the data derivation logic of all original parameter class groups, and obtain the analysis and derivation results of each first-level indicator, second-level indicator, and third-level indicator in the exclusive financial concept knowledge graph;

[0177] The analytical derivation results of all the first-level indicators, second-level indicators, and third-level indicators in the exclusive financial concept knowledge graph are bound to the corresponding nodes in the exclusive financial concept knowledge graph to obtain the data graph based on the financial analysis of the enterprise to be analyzed.

[0178] In this embodiment, starting from each original parameter class group in the exclusive financial concept knowledge graph of the enterprise to be analyzed, all associations in the financial concept knowledge graph are reversely deduced to obtain the data deduction logic of each original parameter class group, including:

[0179] Taking each original parameter class group in the exclusive financial concept knowledge graph of the enterprise to be analyzed as a starting point, all association relationships existing in the financial concept knowledge graph are sequentially connected in the order of the second to last layer, the third to last layer, and so on of the tree structure of the financial concept knowledge graph, to obtain multiple paths starting from each original parameter class group in the exclusive financial concept knowledge graph of the enterprise to be analyzed, which is the data derivation logic of the original parameter class group;

[0180] Among them, the path is composed of primary indicators, secondary indicators, tertiary indicators and original parameter groups connected in series by multiple association relationships.

[0181] In this embodiment, the data derivation logic of the original parameter group represents the calculation process of using the original parameter group to sequentially calculate the specific values ​​of the primary index, the secondary index, and the tertiary index.

[0182] In this embodiment, all association relationships in the financial concept knowledge graph refer not only to the association relationships between the primary indicators, the secondary indicators, and the tertiary indicators, but also to the additional calculation association lines.

[0183] In this embodiment, the single-threaded data deduction logic means that only one primary indicator, secondary indicator, tertiary indicator or specific value can be calculated using each original parameter group or intermediate quantity in the data deduction logic, that is, all nodes in the corresponding path can only lead to one sub-path.

[0184] In this embodiment, the multi-threaded data deduction logic refers to the use of each original parameter class group or intermediate quantity in the data deduction logic to calculate the specific values ​​of two or more first-level indicators and / or second-level indicators and / or third-level indicators, that is, there may be a node in the corresponding path that leads to at least two sub-paths.

[0185] In this embodiment, based on the data derivation logic of all original parameter class groups, the effective financial data of the enterprise to be analyzed is parsed and deduced, and the parsing and derivation results of each first-level indicator, second-level indicator, and third-level indicator in the exclusive financial concept knowledge graph are obtained, including:

[0186] Based on the calculation method of the association relationship tags contained in the data deduction logic of all original parameter class groups, the effective financial data of the enterprise to be analyzed are extracted and calculated in turn, and the analytical deduction results of each first-level indicator, second-level indicator, and third-level indicator in the exclusive financial concept knowledge graph are obtained.

[0187] The beneficial effects of the above technology are: reverse deduction of all associations in the financial concept knowledge graph, obtaining the data deduction logic of each original parameter class group, providing a basis for subsequent analysis and deduction of the effective financial data of the company to be analyzed, and obtaining the analytical deduction results of each first-level indicator, second-level indicator, and third-level indicator in the exclusive financial concept knowledge graph and the data graph based on the financial analysis of the company to be analyzed.

[0188] Example 9

[0189] Based on Example 1, the decision suggestion module refers to Figure 5 ,include:

[0190] The derivation path fitting submodule is used to fit the decision derivation path in the financial analysis basis data map of the enterprise to be analyzed based on the decision logic corresponding to the financial analysis decision purpose;

[0191] The data re-derivation submodule is used to re-derive the node data in the data map of the financial analysis basis of the analyzed enterprise based on the decision derivation path, and obtain the final decision basis for the financial analysis decision purpose;

[0192] The decision suggestion acquisition submodule is used to obtain the decision suggestion corresponding to the financial analysis decision purpose based on the final decision basis of the financial analysis decision purpose.

[0193] In this embodiment, the decision logic corresponding to the financial analysis decision purpose is determined by the decision rules input by the user, which includes which indicators of which data categories (including primary indicators and / or secondary indicators and / or tertiary indicators) in the financial data of the enterprise to be analyzed are used to calculate the final decision basis for the financial analysis decision purpose, and the method logic for obtaining the decision result based on the final decision basis for the financial analysis decision purpose and the decision rules.

[0194] In this embodiment, based on the decision logic corresponding to the financial analysis decision purpose, a decision derivation path is fitted in the financial analysis basis data map of the enterprise to be analyzed, namely:

[0195] Utilizing the decision logic corresponding to the financial analysis decision purpose, the original parameter classes and / or primary indicators and / or secondary indicators and / or tertiary indicators to be used in the decision logic are determined in sequence in the data map based on the financial analysis of the enterprise to be analyzed, and based on the calculation process in the decision logic, the original parameter classes and / or primary indicators and / or secondary indicators and / or tertiary indicators to be used are connected in sequence to obtain a decision derivation path.

[0196] In this embodiment, the node data in the data map of the financial analysis basis of the enterprise to be analyzed is re-derived based on the decision derivation path to obtain the final decision basis for the financial analysis decision purpose, including:

[0197] Determine the calculation process included in the decision derivation path and not included in all the associations of the data map based on the financial analysis of the enterprise to be analyzed as the calculation process to be executed;

[0198] Based on the calculation process to be executed included in the decision derivation path, the node data corresponding to the original parameter class and / or the first-level index and / or the second-level index and / or the third-level index used in the calculation process to be executed are recalculated until the node data of the last node in the decision derivation path is obtained, and the node data of the last node is used as the final decision basis for the financial analysis decision purpose;

[0199] The node data is the performance value of the original parameter class or the first-level indicator or the second-level indicator or the third-level indicator.

[0200] In this embodiment, a decision suggestion corresponding to the financial analysis decision purpose is obtained based on the final decision basis of the financial analysis decision purpose, that is, a decision suggestion corresponding to the financial analysis decision purpose is determined based on the final decision basis contained in the decision logic and the final decision basis of the financial analysis decision purpose, for example, the decision suggestion gives a risk type that needs to be warned.

[0201] The beneficial effects of the above technology are: based on the decision logic corresponding to the financial analysis decision purpose, a decision deduction path is fitted in the financial analysis basis data map of the enterprise to be analyzed, and based on the decision deduction path, the node data in the financial analysis basis data map of the enterprise to be analyzed is re-derived to obtain the final decision basis for the decision recommendation, and finally a reasonable decision recommendation is given.

[0202] Embodiment 10:

[0203] The present invention provides an AI-based automated enterprise financial analysis method, which is applied to any one of the AI-based automated enterprise financial analysis systems described in embodiments 1 to 9, with reference to Figure 6 ,include:

[0204] S1: Obtain valid financial data of the company to be analyzed based on AI;

[0205] S2: Constructing a knowledge graph of financial concepts;

[0206] S3: Analyze the effective financial data of the enterprise to be analyzed based on the financial concept knowledge graph to obtain the data graph based on the financial analysis of the enterprise to be analyzed;

[0207] S4: Based on the decision-making purpose of financial analysis and the data map of the financial analysis basis of the enterprise to be analyzed, obtain decision recommendations corresponding to the decision-making purpose of financial analysis.

[0208] The beneficial effects of the above technologies are: based on AI, effective financial data of the company to be analyzed is obtained, and the financial data is effectively standardized and converted, ensuring the quality and effectiveness of the financial data; and a knowledge graph is introduced to construct a financial concept knowledge graph, and based on the constructed financial concept knowledge graph, the company's financial data is fully automated for in-depth analysis and intelligent decision support. Compared with traditional financial management tools, it does not rely solely on fixed analysis models, but improves the flexibility and scalability of data analysis in the financial analysis process, and can achieve in-depth business insights, thereby providing timely and accurate financial analysis conclusions and decision support in a fully automatic mode, improving the efficiency of corporate financial analysis, the quality of data collection, and the flexibility and depth of data analysis.

[0209] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An AI-based automated enterprise financial analysis system, characterized by: include: Data acquisition module, used to obtain valid financial data of the enterprise to be analyzed based on AI; Knowledge graph construction module, used to build financial concept knowledge graph; The data analysis module is used to analyze the effective financial data of the enterprise to be analyzed based on the financial concept knowledge map, and obtain the data map based on the financial analysis of the enterprise to be analyzed; The decision-making suggestion module is used to obtain decision suggestions corresponding to the financial analysis decision-making purpose based on the financial analysis decision-making purpose and the financial analysis basis data map of the enterprise to be analyzed; Among them, the data analysis module includes: The final indicator screening submodule is used to treat all the third-level indicators in the financial concept knowledge graph that do not have another third-level indicator as the corresponding intermediate calculation amount as the real final-level indicators; The original parameter class selection submodule is used to obtain all the original parameter classes of each real final-level indicator in the enterprise to be analyzed based on the final basis calculation method of each real final-level indicator; The graph supplement submodule is used to compile all original parameter classes of each real final-level indicator into the financial concept knowledge graph as subnodes of the corresponding node in the financial concept knowledge graph, so as to obtain a refined financial concept knowledge graph of the enterprise to be analyzed; The graph simplification submodule is used to merge the same original parameter classes in the refined financial concept knowledge graph of the enterprise to be analyzed to obtain the exclusive financial concept knowledge graph of the enterprise to be analyzed; The data analysis and derivation submodule is used to analyze and deduce the effective financial data of the enterprise to be analyzed based on the exclusive financial concept knowledge map of the enterprise to be analyzed, and obtain the data map based on the financial analysis of the enterprise to be analyzed; Among them, the original parameter class selects submodules, including: An application feasibility evaluation unit is used to determine all possible calculation methods for each third-level indicator and evaluate the application feasibility of each possible calculation method for each third-level indicator for the enterprise to be analyzed; The inspection object distinguishing unit is used to summarize the real final-level indicator with another third-level indicator as the corresponding intermediate calculation amount and the third-level indicator corresponding to the corresponding intermediate calculation amount as the first inspection object, and to take each third-level indicator among all the third-level indicators except all the third-level indicators included in the first inspection object as the second inspection object; A first calculation method screening unit is used to screen out the final basis calculation methods corresponding to the two third-level indicators in each first investigation object from all the basis calculation methods of the two third-level indicators in each first investigation object based on the feasibility of each basis calculation method of each third-level indicator for the enterprise to be analyzed; A second calculation method screening unit is used to use the calculation method corresponding to the maximum application feasibility among all the calculation methods that can be used for each second investigation object as the final calculation method for the third-level indicator corresponding to each second investigation object; The original parameter class selection unit is used to treat all the original parameter classes required in the final basis calculation method of each real final level indicator as all the original parameter classes corresponding to the real final level indicator; The application feasibility assessment unit includes: The calculation method determination subunit is used to determine all the available calculation methods for each third-level indicator; The similar enterprise screening subunit is used to screen out all reference enterprises whose scale similarity with the enterprise to be analyzed is not less than the similarity threshold in the reference enterprise financial analysis database as all enterprises with similar scales to the enterprise to be analyzed; The application feasibility calculation subunit is used to take the ratio of the total application times of each available calculation method of each third-level indicator in all enterprises of similar size to the enterprise to be analyzed and the total application times of all available calculation methods of the corresponding third-level indicator in all enterprises of similar size to the enterprise to be analyzed, as the application feasibility degree of each available calculation method of each third-level indicator for the enterprise to be analyzed.

2. The AI-based automated enterprise financial analysis system according to claim 1, characterized in that: Data acquisition module, including: The financial data collection submodule is used to collect the financial statements, transaction records, and market data of the enterprise to be analyzed and summarize them as the financial data of the enterprise to be analyzed; The data cleaning and verification submodule is used to clean and verify the financial data of the company to be analyzed based on AI to obtain valid financial data of the company to be analyzed.

3. The AI-based automated enterprise financial analysis system according to claim 1, characterized in that: Knowledge graph building blocks include: The indicator naming submodule is used to treat asset quality, income quality, profit quality, cash flow quality and labor efficiency quality as first-level indicators, and determine all measurement dimensions of each first-level indicator as all second-level indicators of each first-level indicator, and all specific measurement indicators of each measurement dimension as all third-level indicators of each second-level indicator; The tree structure building submodule is used to treat all the first-level indicators as all the root nodes of the tree structure, and at the same time, treat all the second-level indicators corresponding to each first-level indicator as all the child nodes of the corresponding root node in the corresponding tree structure, and at the same time, treat all the third-level indicators of each second-level indicator as all the leaf nodes of the corresponding child nodes in the corresponding tree structure, so as to obtain at least one financial concept tree structure map; The association line supplement submodule is used to set an additional calculation association line between the corresponding two third-level indicators to obtain a financial concept knowledge graph when it is determined that there is at least one third-level indicator in all financial concept tree structure graphs that is an intermediate calculation quantity of another third-level indicator.

4. The AI-based automated enterprise financial analysis system according to claim 1, characterized in that: The first calculation method screening unit includes: The application feasibility calculation subunit is used to determine all interrelated calculation method groups among all available calculation methods of the two third-level indicators in the first investigation object, and calculate the actual application feasibility of each interrelated calculation method group based on the application feasibility of each available calculation method of the third-level indicators for the enterprise to be analyzed; The final calculation method screening subunit is used to use the two available calculation methods contained in the interrelated calculation method group with the greatest practical application feasibility among all interrelated calculation method groups of each first investigation object as the final basis calculation methods for the two third-level indicators in each first investigation object.

5. The AI-based automated enterprise financial analysis system according to claim 1, characterized in that: Data analysis and derivation submodule, including: The original parameter class group determination unit is used to group all the original parameter classes of each real final-level indicator in the exclusive financial concept knowledge graph of the enterprise to be analyzed to obtain multiple original parameter class groups; A tree structure reverse derivation unit is used to reversely deduce all associations in the financial concept knowledge graph with each original parameter class group in the exclusive financial concept knowledge graph of the enterprise to be analyzed as a starting point, and obtain the data derivation logic of each original parameter class group, wherein the data derivation logic is a single-threaded data derivation logic or a multi-threaded data derivation logic; The data analysis and derivation unit is used to analyze and derive the effective financial data of the enterprise to be analyzed based on the data derivation logic of all original parameter class groups, and obtain the analysis and derivation results of each first-level indicator, second-level indicator, and third-level indicator in the exclusive financial concept knowledge graph; The analytical derivation results of all the first-level indicators, second-level indicators, and third-level indicators in the exclusive financial concept knowledge graph are bound to the corresponding nodes in the exclusive financial concept knowledge graph to obtain the data graph based on the financial analysis of the enterprise to be analyzed.

6. The AI-based automated enterprise financial analysis system according to claim 1, characterized in that: Decision-making recommendation module, including: The derivation path fitting submodule is used to fit the decision derivation path in the financial analysis basis data map of the enterprise to be analyzed based on the decision logic corresponding to the financial analysis decision purpose; The data re-derivation submodule is used to re-derive the node data in the data map of the financial analysis basis of the analyzed enterprise based on the decision derivation path, and obtain the final decision basis for the financial analysis decision purpose; The decision suggestion acquisition submodule is used to obtain the decision suggestion corresponding to the financial analysis decision purpose based on the final decision basis of the financial analysis decision purpose.

7. The AI-based automated enterprise financial analysis method is characterized by: The AI-based automated enterprise financial analysis system applied to any one of claims 1 to 6 comprises: S1: Obtain valid financial data of the company to be analyzed based on AI; S2: Constructing a knowledge graph of financial concepts; S3: Analyze the effective financial data of the enterprise to be analyzed based on the financial concept knowledge graph to obtain the data graph based on the financial analysis of the enterprise to be analyzed; S4: Based on the decision-making purpose of financial analysis and the data map of the financial analysis basis of the enterprise to be analyzed, obtain decision recommendations corresponding to the decision-making purpose of financial analysis.

Images