Financial report analysis method and system based on natural language processing

Abstract

The present invention discloses a financial report analysis method and system based on natural language processing, and relates to the technical field of financial data processing. The present invention comprises: a table splitting unit splits a project total report and an indicator total report to form a plurality of project reports and indicator reports; a table reorganization unit: it reorganizes into a permission project-indicator report according to the authority of the browsing user. The present invention uses a language conversion unit to convert a target language into a financial language, and uploads it to a processor; a table forming unit forms a project total report and an indicator total report from the financial language; a table splitting unit splits a project total report and an indicator total report to form a plurality of project reports and indicator reports; a table reorganization unit reorganizes into a permission project-indicator report according to the authority of the browsing user, and improves the security and privacy of the table by identifying and reorganizing the target language required in the table, splitting and reorganizing the table, and combining the authority of the browsing user to reorganize into a permission project-indicator report.

Description

Technical Field

[0001] The present invention belongs to the technical field of financial data processing, and in particular, relates to a financial report analysis method and system based on natural language processing. Background Art

[0002] Financial statement analysis is the processing, analysis, comparison, evaluation and interpretation of the data provided by the company's financial statements. Through financial statement analysis, we can determine the company's financial status and diagnose the gains and losses of the company's operation and management. Through analysis, we can determine the company's financial status, operation and management status, business prospects, etc.

[0003] Financial analysis usually refers to an operation and management activity that uses a series of specialized analysis techniques and methods to analyze and evaluate the profitability, operating capacity, debt-paying capacity and growth capacity of the business activities, investment activities and financing activities of the enterprise in the past or present, based on accounting statements and other relevant information, and finally draws conclusions from a financial perspective and puts forward rationalization suggestions. For example, Chinese patent CN110222336A involves the field of big data and provides a financial report analysis method, device, computer equipment and storage medium. Text recognition is performed based on financial reports to obtain original financial business data. From the original financial business data, useful information is identified and extracted for data processing, including a series of data processing processes such as identifying indicators, matching target analysis fields, and calculating target analysis values. The information text used to describe the financial report is obtained as the analysis result, which can realize the analysis of financial reports more quickly, accurately and effectively, thereby improving the analysis efficiency of financial reports.

[0004] However, there is no good solution for the analysis and authority management of financial statements. Therefore, it is necessary to provide a financial report analysis method and system based on natural language processing. Summary of the invention

[0005] The purpose of the present invention is to provide a financial report analysis method and system based on natural language processing, which solves the problems in the background technology by identifying and reorganizing the target language required in the logarithmic table, splitting and reorganizing the logarithmic table, and combining the browsing user's permissions to reorganize it into a permission item-indicator report.

[0006] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0007] As one aspect of the present invention, the present invention is a financial report analysis system based on natural language processing, comprising:

[0008] Language conversion unit: used to convert the target language into financial language and upload it to the processor;

[0009] Data table forming unit: it converts financial language into project summary report and indicator summary report;

[0010] Data table splitting unit: it splits the total project report and the total indicator report into several project reports and indicator reports;

[0011] Data table reorganization unit: It is reorganized into permission item-indicator report according to the permissions of browsing users.

[0012] Furthermore, it also includes a language extraction unit, which is used to extract the target language from the natural language and upload it to the processor, and the language extraction unit retrieves the target language from the processor.

[0013] Furthermore, the language extraction unit performs the following algorithm when extracting the target language from the natural language:

[0014] Receiving natural language, extracting the natural language using a preset target language extraction rule library to obtain an initial language group, wherein the initial language group includes a preset target language;

[0015] S1: Sort the target languages ​​in the extracted initial language group according to the time sequence of the received natural languages;

[0016] S2: Get the target language ranked first and mark it as the top language;

[0017] S3: Obtain all target language combination trees containing the top languages ​​from the target language extraction rule base, and mark them as target tree group 1;

[0018] S4: Obtain the target language ranked second, obtain all target language combination trees containing it from target tree group one, and mark them as target tree group two;

[0019] S5: The same process is repeated until a final target tree is obtained, wherein the final target tree is a target tree containing only one target language.

[0020] The remaining target languages ​​are re-arranged in chronological order, and steps S2 to S5 are repeated until all target languages ​​are analyzed.

[0021] Furthermore, when the language extraction unit extracts the target language from the natural language:

[0022] When the extracted target language fails to match the target language combination tree, receiving an isolated data instruction transmitted by the target language extraction rule base;

[0023] All isolated data instructions are transmitted to the user end through the processor, and the user uploads a new target language combination tree or deletes the target language corresponding to the isolated data instruction through the user end.

[0024] Furthermore, the language conversion unit converts the target language into the financial language by:

[0025] Choose a final target tree;

[0026] Obtain all target languages ​​in the final target tree, and split the target languages ​​into data languages ​​and text languages;

[0027] Compare the text language with the financial professional language in the database one by one:

[0028] When the comparison is consistent, the text language is directly converted into financial language;

[0029] When the comparison is inconsistent, the text language and its corresponding data language are listed as questionable languages ​​and transmitted to the processor, and then the language extraction unit extracts the target language from the natural language and re-analyzes the questionable language;

[0030] Until all target language conversions are completed.

[0031] Furthermore, the step of reanalyzing the question language is:

[0032] F1: Get the final target tree where the question language is located;

[0033] F2: Obtain the total number of characters Z in the text language of the question language, the number of characters Q that are different from the financial professional language, and calculate the difference ratio, which is Q / Z;

[0034] F3: Obtain the difference ratio between the text language in the question language and the target language in the final target tree according to F2;

[0035] If both difference ratios are smaller than the preset value C1, the questionable language is transmitted to the user end via the processor for verification.

[0036] Furthermore, the rule for the table forming unit to form a project summary report and an indicator summary report from the financial language is:

[0037] Each target language in the final target tree forms a report. If the target language contains "project", the word "project" is given to the header of the report. If the target language contains "indicator", the word "indicator" is given to the header of the report.

[0038] Get the header of each report, and combine the items in the header + the reports with the same name into a project summary report, and combine the indicators in the header + the reports with the same name into an indicator summary report.

[0039] Furthermore, when the data table splitting unit splits the project total report and the indicator total report into a plurality of project reports and indicator reports:

[0040] Split the project summary report into 3 project reports: income data table, expenditure data table, and profit data table;

[0041] Split the indicator summary report into 3 indicator reports: short-term indicator data table, medium-term indicator data table, and long-term indicator data table;

[0042] When the project summary report and the indicator summary report are split to form several project reports and indicator reports, permission identification is required when browsing.

[0043] Furthermore, the permission identification steps are as follows:

[0044] Perform minimum permission division on each report to form multi-level permissions corresponding to each report;

[0045] Authenticate the identity of the browsing user, and query and allocate permission thresholds at the system administrator;

[0046] The system administrator assigns a permission level to the browsing user:

[0047] If the permission threshold ≧ E1, it is divided into first-level permissions;

[0048] If E2 ≦ permission threshold < E1, it is divided into second-level permissions;

[0049] And so on, a total of e-level permissions are divided. Each level of permission can query at least one data table, and the data tables corresponding to each level of permission are different;

[0050] Package all the data tables corresponding to each level of permission and reorganize them to form a new permission project - indicator report.

[0051] The financial report analysis method based on natural language processing specifically includes the following steps:

[0052] Step 1: Convert the target language into financial language through the language conversion unit and upload it to the processor;

[0053] Step 2: Form a project summary report and an indicator summary report from the financial language through the data table formation unit;

[0054] Step 3: Split the project summary report and the indicator summary report into several project reports and indicator reports through the data table splitting unit;

[0055] Step 4: Reorganize according to the permissions of the browsing user into a permission project - indicator report through the data table reorganization unit.

[0056] The present invention has the following beneficial effects:

[0057] The present invention uses a language conversion unit to convert a target language into a financial language and upload it to a processor; a table forming unit forms a project summary report and an indicator summary report from the financial language; a table splitting unit splits the project summary report and the indicator summary report into a plurality of project reports and indicator reports; a table reorganization unit reorganizes the table into an authority project-indicator report according to the authority of the browsing user, identifies and reorganizes the target language required in the table, splits and reorganizes the table, and reorganizes the table into an authority project-indicator report in combination with the authority of the browsing user, thereby improving the security and privacy of the table.

[0058] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for describing the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 It is a structural schematic diagram of the financial report analysis system based on natural language processing of the present invention;

[0061] Figure 2 Schematic diagram of the target language combination tree structure of an embodiment of the present invention. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] See also Figure 1 and 2 As shown:

[0064] Embodiment 1:

[0065] As an embodiment of the present invention, the present invention is a financial report analysis system based on natural language processing, comprising:

[0066] Language conversion unit: used to convert the target language into financial language and upload it to the processor; as an embodiment provided by the present invention, it also includes a language extraction unit, which is used to extract the target language from the natural language and upload it to the processor, and the language extraction unit retrieves the target language from the processor.

[0067] As an embodiment provided by the present invention, the language extraction unit executes the following algorithm when extracting the target language from the natural language:

[0068] Receiving natural language, extracting the natural language using a preset target language extraction rule library to obtain an initial language group, wherein the initial language group includes a preset target language;

[0069] S1: Sort the target languages ​​in the extracted initial language group according to the time sequence of the received natural languages;

[0070] S2: Get the target language ranked first and mark it as the top language;

[0071] S3: Obtain all target language combination trees containing the top languages ​​from the target language extraction rule base, and mark them as target tree group 1; Figure 2 As shown, it is a target language combination tree including the target language "G project";

[0072] S4: Obtain the target language ranked second, obtain all target language combination trees containing it from target tree group one, and mark them as target tree group two;

[0073] S5: The same process is repeated until a final target tree is obtained, wherein the final target tree is a target tree containing only one target language.

[0074] The remaining target languages ​​are re-arranged in chronological order, and steps S2 to S5 are repeated until all target languages ​​are analyzed.

[0075] As an embodiment provided by the present invention, when the language extraction unit extracts the target language from the natural language:

[0076] When the extracted target language fails to match the target language combination tree, receiving an isolated data instruction transmitted by the target language extraction rule base;

[0077] All isolated data instructions are transmitted to the user end through the processor, and the user uploads a new target language combination tree or deletes the target language corresponding to the isolated data instruction through the user end.

[0078] Data table forming unit: it converts financial language into project summary report and indicator summary report;

[0079] Data table splitting unit: it splits the total project report and the total indicator report into several project reports and indicator reports;

[0080] Data table reorganization unit: It is reorganized into permission item-indicator report according to the permissions of browsing users.

[0081] As an embodiment provided by the present invention, the steps of converting the target language into financial language by the language conversion unit are as follows:

[0082] Choose a final target tree;

[0083] Obtain all target languages ​​in the final target tree, and split the target languages ​​into data languages ​​and text languages;

[0084] Compare the text language with the financial professional language in the database one by one:

[0085] When the comparison is consistent, the text language is directly converted into financial language;

[0086] When the comparison is inconsistent, the text language and its corresponding data language are listed as questionable languages ​​and transmitted to the processor, and then the language extraction unit extracts the target language from the natural language and re-analyzes the questionable language;

[0087] Until all target language conversions are completed.

[0088] As an embodiment provided by the present invention, the step of reanalyzing the question language is:

[0089] F1: Get the final target tree where the question language is located;

[0090] F2: Obtain the total number of characters Z in the text language of the question language, the number of characters Q that are different from the financial professional language, and calculate the difference ratio, which is Q / Z;

[0091] F3: Obtain the difference ratio between the text language in the question language and the target language in the final target tree according to F2;

[0092] If both difference ratios are smaller than the preset value C1, the questionable language is transmitted to the user end via the processor for verification.

[0093] As an embodiment provided by the present invention, the rule for the table forming unit to form a project summary report and an indicator summary report from financial language is as follows:

[0094] Each target language in the final target tree forms a report. If the target language contains "project", the word "project" is given to the header of the report. If the target language contains "indicator", the word "indicator" is given to the header of the report.

[0095] Get the header of each report, and combine the items in the header + the reports with the same name into a project summary report, and combine the indicators in the header + the reports with the same name into an indicator summary report.

[0096] As an embodiment provided by the present invention, when the data table splitting unit splits the project total report and the indicator total report into several project reports and indicator reports:

[0097] Split the project summary report into 3 project reports: income data table, expenditure data table, and profit data table;

[0098] Split the indicator summary report into 3 indicator reports: short-term indicator data table, medium-term indicator data table, and long-term indicator data table;

[0099] When the project summary report and the indicator summary report are split to form several project reports and indicator reports, permission identification is required during browsing.

[0100] As an embodiment provided by the present invention, the permission identification steps for the browsing user are as follows:

[0101] Perform minimum permission division on each report to form multi-level permissions corresponding to each report;

[0102] Authenticate the identity identifier of the browsing user and query the assigned permission threshold at the system administrator;

[0103] The system administrator assigns a permission level to the browsing user:

[0104] If the permission threshold ≥ E1, it is divided into first-level permissions;

[0105] If E2 ≤ permission threshold < E1, it is divided into second-level permissions;

[0106] And so on, a total of e levels of permissions are divided. Each level of permission can query at least one data table, and the data tables corresponding to each level of permission are different;

[0107] Package all the data tables corresponding to each level of permission and reorganize them to form a new permission project - indicator report.

[0108] Embodiment 2:

[0109] As another aspect of the present invention, a financial report analysis method based on natural language processing specifically includes the following steps:

[0110] Step 1: Convert the target language into financial language through the language conversion unit and upload it to the processor;

[0111] Step 2: Form a project summary report and an indicator summary report from the financial language through the data table formation unit;

[0112] Step 3: Split the project summary report and the indicator summary report into several project reports and indicator reports through the data table splitting unit;

[0113] Step 4: Reorganize them into a permission project - indicator report according to the permissions of the browsing user through the data table reorganization unit.

[0114] The financial report analysis method and system based on natural language processing are used to convert the target language into financial language through a language conversion unit and upload it to a processor; the table formation unit forms a project summary report and an indicator summary report from the financial language; the table splitting unit splits the project summary report and the indicator summary report into several project reports and indicator reports; the table reorganization unit reorganizes the table into a permission project-indicator report according to the permissions of the browsing user, identifies and reorganizes the target language required in the table, splits and reorganizes the table, and reorganizes it into a permission project-indicator report in combination with the permissions of the browsing user, thereby improving the security and privacy of the table.

[0115] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0116] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A financial report analysis system based on natural language processing, characterized in that: include: Language conversion unit: used to convert the target language into financial language and upload it to the processor; Data table forming unit: it converts financial language into project summary report and indicator summary report; Data table splitting unit: it splits the total project report and the total indicator report into several project reports and indicator reports; Data table reorganization unit: it is reorganized into permission item-indicator report according to the permissions of browsing users; It also includes a language extraction unit, which is used to extract the target language from the natural language and upload it to the processor, and the language conversion unit retrieves the target language from the processor; The language extraction unit performs the following algorithm when extracting the target language from the natural language: Receiving natural language, extracting the natural language using a preset target language extraction rule library to obtain an initial language group, wherein the initial language group includes a preset target language; S1: Sort the target languages ​​in the extracted initial language group according to the time sequence of the received natural languages; S2: Get the target language ranked first and mark it as the top language; S3: Obtain all target language combination trees containing top languages ​​from the target language extraction rule base, and mark them as target tree group 1; S4: Obtain the target language ranked second, obtain all target language combination trees containing it from target tree group one, and mark them as target tree group two; S5: The same process is repeated until a final target tree is obtained, wherein the final target tree is a target tree containing only one target language. The remaining target languages ​​are re-arranged in chronological order, and steps S2 to S5 are repeated until all target languages ​​are analyzed.

2. The financial report analysis system based on natural language processing according to claim 1, characterized in that: When the language extraction unit extracts the target language from the natural language: When the extracted target language fails to match the target language combination tree, receiving an isolated data instruction transmitted by the target language extraction rule base; All isolated data instructions are transmitted to the user end through the processor, and the user uploads a new target language combination tree or deletes the target language corresponding to the isolated data instruction through the user end.

3. The financial report analysis system based on natural language processing according to claim 2 is characterized in that: The steps of converting the target language into financial language by the language conversion unit are as follows: Choose a final target tree; Obtain the target language in the final target tree, and split the target language into data language and text language; Compare the text language with the financial professional language in the database one by one: When the comparison is consistent, the text language is directly converted into financial language; When the comparison is inconsistent, the text language and its corresponding data language are listed as questionable languages ​​and transmitted to the processor, and then the language extraction unit extracts the target language from the natural language and re-analyzes the questionable language; Until all target language conversions are completed.

4. The financial report analysis system based on natural language processing according to claim 3 is characterized in that: The steps of reanalyzing the question language are as follows: F1: Get the final target tree where the question language is located; F2: Obtain the total number of characters Z in the text language of the question language and the number of different characters Q in the financial professional language, and calculate the difference ratio, which is Q / Z; F3: Obtain the difference ratio between the text language in the question language and the target language in the final target tree; If both difference ratios are smaller than the preset value C1, the questionable language is transmitted to the user end via the processor for verification.

5. The financial report analysis system based on natural language processing according to claim 4 is characterized in that: The rules for the table forming unit to form the project total report and the indicator total report from the financial language are as follows: The target language in each end - goal tree forms a report. If the target language contains: "project", the word "project" is given to its header; if the target language contains: "indicator", the word "indicator" is given to its header. Obtain the headers in each report. Reports with the same project and name in the headers form a total project report, and reports with the same indicator and name in the headers form a total indicator report.

6. The financial report analysis system based on natural language processing according to claim 5 is characterized in that: When the data table splitting unit splits the total project report and the total indicator report into several project reports and indicator reports: The total project report is split into 3 project reports: income data table, expenditure data table, and profit data table. The total indicator report is split into 3 indicator reports: short - term indicator data table, medium - term indicator data table, and long - term indicator data table. When the total project report and the total indicator report are split into several project reports and indicator reports, permission identification is required during browsing.

7. The financial report analysis system based on natural language processing according to claim 6 is characterized in that: The permission identification steps are as follows: Perform the minimum - level permission division for each report to form multi - level permissions corresponding to each report. Authenticate the identity identifier of the browsing user and query the assigned permission threshold at the system administrator. The system administrator assigns a permission level to the browsing user: If the permission threshold ≧ E1, it is divided into first - level permissions. If E2 ≦ permission threshold < E1, it is divided into second - level permissions. And so on, a total of e - level permissions are divided. Each level of permission can query at least one data table respectively, and the data tables corresponding to each level of permission are different. Package all the data tables corresponding to each level of permission and reorganize them to form a new permission project - indicator report.

8. A financial report analysis method based on natural language processing, characterized in that: The method described is based on the analysis system described in any one of claims 1 - 7, and specifically includes the following steps: Step 1: Convert the target language into financial language through the language conversion unit and upload it to the processor. Step 2: Form a total project report and a total indicator report from the financial language through the data table forming unit. Step 3: Split the total project report and the total indicator report into several project reports and indicator reports through the data table splitting unit. Step 4: Reorganize them into a permission project - indicator report according to the permissions of the browsing user through the data table reorganization unit.

Images