CN117216801A - Enterprise financial data safety management system and method based on artificial intelligence - Google Patents

Abstract

The invention discloses an enterprise financial data security management system and method based on artificial intelligence, and relates to the technical field of enterprise financial security management. The invention provides an enterprise financial data security management system and method based on artificial intelligence. By collecting enterprise financial After the collection of data samples is processed and features are extracted, a deep learning model is established and trained to obtain the comprehensive classification coefficient Zhxs and compare it with the standard threshold to determine the risk level and perform corresponding encryption processing. Different security measures are adopted for different levels of risk data. , to prevent unauthorized access and data leakage; by establishing a deep learning model and analyzing the behavior patterns of encrypted risk data, abnormal activities and access intrusion behaviors can be monitored in real time. Use artificial intelligence technology to identify risky behaviors and reduce possible false positives and negatives in anomaly detection and risk assessment in financial data security management systems.

Description

An artificial intelligence-based enterprise financial data security management system and method

Technical field

The invention relates to the technical field of enterprise financial security management, specifically an enterprise financial data security management system and method based on artificial intelligence.

Background technique

The enterprise financial data security management system exists to solve the financial data security and management challenges faced by enterprises and provide a comprehensive and effective solution. Financial data is one of the most important and sensitive assets of a business. Securing financial data is critical to the sustainable development of your business.

The existing enterprise financial data security management system simply uploads the enterprise's financial data to the system and then transmits, accesses and modifies it with some permissions of the managers. However, for a large amount of financial data, manual management, submission, storage and access processes are required. The encryption behavior performed in the financial data security management system is relatively not intelligent enough, and manual management can easily lead to false positives and false negatives in anomaly detection and risk assessment in the financial data security management system.

Contents of the invention

(1) Technical problems solved

In view of the shortcomings of the existing technology, the present invention provides an artificial intelligence-based enterprise financial data security management system and method. By collecting a collection of enterprise financial data samples for processing and feature extraction, a deep learning model is established and intelligent model training is performed. , obtain the comprehensive classification coefficient Zhxs, and compare the comprehensive classification coefficient Zhxs with the standard threshold to determine the risk level and perform corresponding encryption processing, which improves the security and protection level of corporate financial data. Different security measures are adopted for different levels of risk data to prevent unauthorized access and data leakage; by establishing a deep learning model and analyzing the behavior patterns of encrypted risk data, abnormal activities and access intrusion behaviors can be monitored in real time. Use artificial intelligence technology to identify risky behaviors and generate corresponding reports and early warning notices to help enterprises discover and respond to potential risks in a timely manner, and reduce possible false positives and negatives in anomaly detection and risk assessment in financial data security management systems.

(2) Technical solutions

In order to achieve the above objectives, the present invention is implemented through the following technical solutions: an artificial intelligence-based enterprise financial data security management method, including the following steps:

S1. Collect and obtain collective samples of corporate financial data, and process the collective samples of corporate financial data;

S2. Extract features from the collected samples of processed financial data to obtain characteristic financial data;

S3. Establish a deep learning model, classify the processed characteristic financial data, conduct model training, and obtain the comprehensive classification coefficient Zhxs;

S4. Intelligent security encryption management compares the comprehensive classification coefficient Zhxs with the standard threshold. If it exceeds the standard threshold, it is determined to be core high-level risk data; if the comprehensive classification coefficient Zhxs is lower than 30% of the standard threshold, it is determined to be medium-level risk data; If the comprehensive classification coefficient Zhxs is lower than 40% of the standard threshold, it is determined to be low-level risk data; if the comprehensive classification coefficient Zhxs is lower than 50% of the standard threshold, it is determined to be public-level data; and the corresponding level risk data is corresponding encryption processing;

S5. Analyze the behavior patterns of the corresponding level of risk data for encryption processing, establish benchmarks, monitor abnormal activities and access intrusion behaviors, identify risk behaviors through artificial intelligence technology, generate corresponding reports, and send early warning notifications.

Preferably, the S1 step includes:

S11. Collect a collection of corporate financial data samples, including classified revenue, expenditure and profit data; it also includes unclassified data, which requires manual labeling of categories;

Revenue includes the company's sales revenue, service fees and interest income; expenses include the company's purchase costs, monthly salary statements, rent and transportation costs; profits include the company's net profit and gross profit; unclassified data includes liabilities and shareholders' equity. , taxation and financial statement classification; liabilities include the company's short-term liabilities, long-term liabilities and bonds; shareholders' equity includes the company's equity information, capital reserves and surplus reserves; taxation includes the company's taxes, tax declarations and tax adjustments; financial statements include The company's balance sheet, income statement and cash flow statement;

S12. Process the collected samples of corporate financial data. The processing includes data cleaning, missing value processing and normalization processing.

Preferably, the S2 step includes extracting features from the set sample of financial data, including numerical feature extraction and text type feature extraction. The numerical feature extraction includes extracting total amount, average value and variance, and the text type feature extraction includes Extract keywords and word frequency information to obtain characteristic financial data;

Total extraction is used to calculate the total of financial data, including calculating total revenue, total expenses, and total profit;

Average extraction is used to calculate the average of financial data, including calculating average sales, evaluation costs, and average profits;

Variance extraction is used to calculate the variance of financial data, measure the degree of change in the data, and obtain a variance change waveform table;

Text type feature extraction is used to process text fields in financial data, segment the text, extract stems, and convert the text into a keyword list. The keywords are set to product names, supplier names, and customer names; word frequency information extraction includes Count the number of times keywords appear in text in financial data, and calculate the word frequency of each keyword in the financial data sample collection as a text type feature.

Preferably, the S3 step includes:

S31. Extract the total amount, average value and variance as numerical features, and extract keywords and word frequency information as text type features; associate these features with the classification labels of the financial data samples;

S3. Establish a deep learning model for classification task training, including one of decision trees, support vector machines and neural networks;

S4. Predict new characteristic financial data samples based on the deep learning model and obtain the classification results; calculate the comprehensive classification coefficient Zhxs based on the classification results.

Preferably, the comprehensive classification coefficient Zhxs is obtained by the following formula:

In the formula, ,/> ,/> ,...,/> It is a feature value extracted from feature financial data, set to be a numerical feature or text feature;/> ,/> ,/> ,...,/> is the weight of the corresponding feature;/> for the correction constant.

Preferably, the S4 includes,

S41. Set standard thresholds for judging different levels of risks; set standard thresholds to 30%, 40% and 50%;

S42. Compare the comprehensive classification coefficient Zhxs of each characteristic financial data sample, and judge the risk level according to different conditions:

If Zhxs exceeds the standard threshold, it is determined to be core high-level risk data;

If Zhxs is 30% lower than the standard threshold, it is determined to be medium-level risk data;

If Zhxs is lower than 40% of the standard threshold, it is determined to be low-level risk data;

If Zhxs is lower than 50% of the standard threshold, it is determined to be public-level data, which means risk-free data;

S43. According to the determination result of the risk level, perform corresponding encryption method processing on the risk data of the corresponding level.

Preferably, the encryption method of S43 is determined according to different levels of risk data; specifically, the encryption processing method includes:

S431. Core high-level risk data: Use three letters plus three digits as the password to encrypt the data, set corresponding access rights and control rights, and perform identity verification and recording based on the access timestamp;

S432, medium-level risk data: Use six pure letters as the password to encrypt the data, and set corresponding access rights and control rights;

S433, low-level risk data: Use a pure six-digit number as the password to encrypt the data, and set corresponding access rights;

S434. No encryption processing is performed, the access permission is opened, but the access records are backed up and recorded.

Preferably, the S5 includes:

S51. Through learning and analyzing normal behavior patterns, establish behavioral pattern benchmarks for each level of risk data; behavioral pattern benchmarks include access frequency, operation type and data transmission mode;

S52. Based on the established behavioral pattern benchmark, use intelligent anomaly monitoring to conduct real-time monitoring of the encrypted level risk data; if abnormalities in access frequency, operation type, and data transmission are detected, it will be determined as a first-level risk behavior;

S53. Use network traffic sensors, log sensors, and authentication sensors to monitor access intrusions, including unauthorized access, abnormal login attempts, and data leaks. If access intrusions are identified, they are determined to be secondary risk behaviors;

S54. Combine the first-level risk behavior and the second-level risk behavior to generate a corresponding risk report. When a risk report is identified, the risk report is analyzed and an early warning command notification is generated and sent to the security team.

Preferably, the early warning command notification includes one or more of email, text message and instant message.

An enterprise financial data security management system based on artificial intelligence, including a data collection module, a data preprocessing module, a feature extraction module, a model training and classification module, a risk level determination module, a backup module, an encryption module, a security control module, and anomaly monitoring modules and early warning modules;

The data collection module collects corporate financial data from different data sources and obtains a collection of corporate financial data samples;

The data preprocessing module is used to clean the collection samples of corporate financial data, remove duplicate values and handle missing values;

The feature extraction module is used to extract key features from financial data, including numerical features and text type features;

The model training and classification module is used to establish a deep learning model or other machine learning model, input the processed characteristic financial data into the model, and calculate and obtain the comprehensive classification coefficient Zhxs;

The risk level determination module is used to compare the comprehensive classification coefficient Zhxs with the preset standard threshold to determine the risk level of financial data. According to the set threshold conditions, the data is divided into different levels of risks, including core high-level risks and medium-level risks. Risk, low-level risk and public-level data;

The backup module is used for periodic backup of data corresponding to core high-level risks and medium-level risks;

The encryption module is used to encrypt risk data of the corresponding level;

The security control module is used to set corresponding access rights and control measures to ensure that only authorized personnel can access and process sensitive data;

The anomaly monitoring module is used to monitor the behavior patterns and access activities of encrypted risk data, identify abnormal behaviors and intrusion behaviors, and generate corresponding reports, including details of abnormal behaviors and risk assessment results;

The early warning module is used to send early warning notifications to relevant responsible persons or security teams based on corresponding reports, so that timely measures can be taken to deal with risk events.

(3) Beneficial effects

The invention provides an enterprise financial data security management system and method based on artificial intelligence. It has the following beneficial effects:

(1) This artificial intelligence-based enterprise financial data security management system and method can comprehensively and accurately express the characteristic information of financial data by extracting numerical features and text type features from a collection of financial data samples. It helps to comprehensively and accurately express the characteristic information of financial data, provide important statistical indicators and trends, and capture key text information and word frequency analysis. By extracting comprehensive features, establishing a deep learning model for classification training, and calculating the comprehensive classification coefficient Zhxs, the classification accuracy and flexibility of financial data can be improved. This helps to better understand and analyze the classification of financial data, providing useful information and basis for subsequent risk assessment, encryption management and behavioral monitoring.

(2) This artificial intelligence-based enterprise financial data security management system and method can improve the level of data security and privacy protection through customized encryption processing for different levels of risks. Strong encryption methods for core high-level risk data can effectively prevent unauthorized access and data leakage, and protect the security of sensitive information. Encryption of medium-level and low-level risk data provides moderate data protection, ensuring that data is not vulnerable to malicious attacks during transmission and storage. Backing up access records can help monitor and track data access behavior and improve data traceability and credibility. The access rights and control rights settings in the encryption processing method make the access and control of different levels of risk data more flexible. Corresponding permission levels can be set according to different business needs and security policies to ensure that only authorized personnel can access and operate the corresponding level of risk data, improving the compliance and security of data access. By customizing encryption of risk data based on risk level determination results, data security, privacy protection and access control capabilities can be improved. This helps protect a business’s financial data from risks and threats and ensures the confidentiality and integrity of data during transmission, storage and processing.

(3) This artificial intelligence-based enterprise financial data security management system and method can monitor and identify risk behaviors, generate corresponding risk reports, and send early warning notifications in a timely manner , improve risk response capabilities and safety protection levels.

(4) This artificial intelligence-based enterprise financial data security management system, through the synergy of each module of the system, the artificial intelligence-based enterprise financial data security management method can provide effective risk management and protection measures, reduce the risk of ensuring financial data Security, integrity and availability, based on artificial intelligence technology, improve intelligence and reduce the problem of false positives and false negatives caused by abnormal monitoring and risk assessment caused by administrators' manual authorization to manage financial data in the system.

Description of drawings

Figure 1 is a schematic diagram of the steps of the enterprise financial data security management method based on artificial intelligence according to the present invention;

Figure 2 is a schematic block diagram of the enterprise financial data security management system based on artificial intelligence according to the present invention;

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

Example 1

The enterprise financial data security management system exists to solve the financial data security and management challenges faced by enterprises and provide a comprehensive and effective solution. Financial data is one of the most important and sensitive assets of a business. Securing financial data is critical to the sustainable development of your business.

The existing enterprise financial data security management system simply uploads the enterprise's financial data to the system and then transmits, accesses and modifies it with some permissions of the managers. However, for a large amount of financial data, manual management, submission, storage and access processes are required. The encryption behavior performed in the financial data security management system is relatively not intelligent enough, and manual management can easily lead to false positives and false negatives in anomaly detection and risk assessment in the financial data security management system.

The present invention provides an artificial intelligence-based enterprise financial data security management method. Please refer to Figure 1, which includes the following steps:

S1. Collect and obtain collective samples of corporate financial data, and process the collective samples of corporate financial data;

S2. Extract features from the collected samples of processed financial data to obtain characteristic financial data;

S3. Establish a deep learning model, classify the processed characteristic financial data, conduct model training, and obtain the comprehensive classification coefficient Zhxs;

S4. Intelligent security encryption management compares the comprehensive classification coefficient Zhxs with the standard threshold. If it exceeds the standard threshold, it is determined to be core high-level risk data; if the comprehensive classification coefficient Zhxs is lower than 30% of the standard threshold, it is determined to be medium-level risk data; If the comprehensive classification coefficient Zhxs is lower than 40% of the standard threshold, it is determined to be low-level risk data; if the comprehensive classification coefficient Zhxs is lower than 50% of the standard threshold, it is determined to be public-level data; and the corresponding level risk data is corresponding encryption processing;

S5. Analyze the behavior patterns of the corresponding level of risk data for encryption processing, establish benchmarks, monitor abnormal activities and access intrusion behaviors, identify risk behaviors through artificial intelligence technology, generate corresponding reports, and send early warning notifications.

In this embodiment, a collection of corporate financial data samples is collected through S1-S3 for processing and feature extraction, then a deep learning model is established, intelligent model training is performed, the comprehensive classification coefficient Zhxs is obtained, and the comprehensive classification coefficient Zhxs and the standard are obtained through S4 Threshold comparison determines the risk level and performs corresponding encryption processing, which improves the security and protection level of corporate financial data. Different security measures are adopted for different levels of risk data to prevent unauthorized access and data leakage; by establishing a deep learning model and analyzing the behavior patterns of encrypted risk data, abnormal activities and access intrusion behaviors can be monitored in real time. Use artificial intelligence technology to identify risky behaviors and generate corresponding reports and early warning notices to help enterprises discover and respond to potential risks in a timely manner, and reduce possible false positives and negatives in anomaly detection and risk assessment in financial data security management systems.

Example 2

This embodiment is an explanation of what was done in Embodiment 1. Specifically, the step S1 includes:

S11. Collect a collection of corporate financial data samples, including classified revenue, expenditure and profit data; it also includes unclassified data, which requires manual labeling of categories;

Revenue includes the company's sales revenue, service fees and interest income; expenses include the company's purchase costs, monthly salary statements, rent and transportation costs; profits include the company's net profit and gross profit; unclassified data includes liabilities and shareholders' equity. , taxation and financial statement classification; liabilities include the company's short-term liabilities, long-term liabilities and bonds; shareholders' equity includes the company's equity information, capital reserves and surplus reserves; taxation includes the company's taxes, tax declarations and tax adjustments; financial statements include The company's balance sheet, income statement and cash flow statement;

S12. Process the collected samples of corporate financial data. The processing includes data cleaning, missing value processing and normalization processing.

In this embodiment, through the data cleaning process, errors, duplications, incomplete or inconsistent data in the data set samples can be removed. This helps improve data accuracy and consistency and avoids erroneous results in subsequent processing and analysis. In financial data, there may be missing values, where some data items are missing values or information. Through missing value processing, appropriate methods can be used to fill in or impute missing data, making the data set sample more complete. This helps improve data availability and reliability. Financial data may have different units of measurement and ranges of values. Through normalization, the data can be converted to a unified scale, making different features comparable. This helps improve the effectiveness of data analysis and model training and reduce biases caused by different scales. Data cleaning, missing value handling, and normalization help improve the quality of financial data. Higher-quality data collection samples can provide more accurate and reliable features for model training and analysis, thereby improving the accuracy and credibility of subsequent risk assessment and decision-making.

Example 3

This embodiment is an explanation of what was done in Embodiment 1. Specifically, step S2 includes extracting features from a collection sample of financial data, including numerical feature extraction and text type feature extraction. The numerical feature extraction includes extracting Total amount, average value and variance, the text type feature extraction includes extracting keywords and word frequency information to obtain characteristic financial data;

Total amount extraction is used to calculate the total amount of financial data, including calculating total revenue, total expenditure, and total profit; through the total amount, average value, and variance in numerical feature extraction, important statistical indicators of financial data can be calculated. Totals reflect the overall picture of revenue, expenses and profits.

Average extraction is used to calculate the average of financial data, including calculating average sales, evaluation costs, and average profits; the average provides the average level of financial data, and the variance measures the degree of variation in the data. These indicators and trends can help companies understand the overall situation and changing trends of financial data.

Variance extraction is used to calculate the variance of financial data, measure the degree of change in the data, and obtain a variance change waveform table;

Text type feature extraction is used to process text fields in financial data, segment the text, extract stems, and convert the text into a keyword list. The keywords are set to product names, supplier names, and customer names; word frequency information extraction includes Count the number of times keywords appear in text in financial data, and calculate the word frequency of each keyword in the financial data sample collection as a text type feature.

In this embodiment, by extracting numerical features and text type features from a collection of financial data samples, the feature information of the financial data can be expressed comprehensively and accurately. It helps to comprehensively and accurately express the characteristic information of financial data, provide important statistical indicators and trends, and capture key text information and word frequency analysis. These features can provide a rich data foundation and useful information for subsequent model training, classification and analysis.

Example 4

This embodiment is an explanation of what was done in Embodiment 3. Specifically, the step S3 includes:

S31. Extract the total amount, average value and variance as numerical features, and extract keywords and word frequency information as text type features; associate these features with the classification labels of financial data samples; feature financial data with rich information can be obtained. Such feature extraction and association help improve the expressive power and classification accuracy of data.

S3. Establish a deep learning model for classification task training, including one of decision tree, support vector machine and neural network; the deep learning model has strong learning and expression capabilities and can effectively learn the patterns and characteristics of financial data. Improve classification accuracy and robustness.

S4. Predict new characteristic financial data samples based on the deep learning model and obtain the classification results; calculate the comprehensive classification coefficient Zhxs based on the classification results. Use the deep learning model to predict new characteristic financial data samples to obtain the classification results, and then calculate the comprehensive classification coefficient Zhxs. The comprehensive classification coefficient combines the weight and eigenvalue of each feature to comprehensively evaluate the classification results of the sample. Such a comprehensive classification coefficient can provide a quantitative indicator to help evaluate and compare the classification degree of different samples.

The comprehensive classification coefficient Zhxs is obtained by the following formula:

In the formula, ,/> ,/> ,...,/> It is a feature value extracted from feature financial data, set to be a numerical feature or text feature;/> ,/> ,/> ,...,/> is the weight of the corresponding feature;/> for the correction constant.

Use the deep learning model to predict new characteristic financial data samples to obtain the classification results, and then calculate the comprehensive classification coefficient Zhxs. The comprehensive classification coefficient combines the weight and eigenvalue of each feature to comprehensively evaluate the classification results of the sample. Such a comprehensive classification coefficient can provide a quantitative indicator to help evaluate and compare the classification degree of different samples.

In this embodiment, by extracting comprehensive features, establishing a deep learning model for classification training, and calculating the comprehensive classification coefficient Zhxs, the classification accuracy and flexibility of financial data can be improved. This helps to better understand and analyze the classification of financial data, providing useful information and basis for subsequent risk assessment, encryption management and behavioral monitoring.

Example 5

This embodiment is an explanation of what was done in Embodiment 1. Specifically, the S4 includes:

S41. Set standard thresholds for judging different levels of risks; set standard thresholds to 30%, 40% and 50%;

S42. Compare the comprehensive classification coefficient Zhxs of each characteristic financial data sample, and judge the risk level according to different conditions:

If Zhxs exceeds the standard threshold, it is determined to be core high-level risk data;

If Zhxs is 30% lower than the standard threshold, it is determined to be medium-level risk data;

If Zhxs is lower than 40% of the standard threshold, it is determined to be low-level risk data;

If Zhxs is lower than 50% of the standard threshold, it is determined to be public-level data, which means risk-free data;

S43. According to the determination result of the risk level, perform corresponding encryption method processing on the risk data of the corresponding level. By setting a standard threshold and comparing the comprehensive classification coefficient Zhxs with the standard threshold, different levels of risk data can be accurately judged. This ensures a strong focus on and protection of core high-level risk data, as well as appropriate processing of mid-level, low-level and public-level data, improving the accuracy and effectiveness of risk management.

The encryption method of S43 is determined according to different levels of risk data; the specific encryption processing methods include:

S431. Core high-level risk data: Use three letters plus three digits as the password to encrypt the data, set corresponding access rights and control rights, and perform identity verification and recording based on the access timestamp;

S432, medium-level risk data: Use six pure letters as the password to encrypt the data, and set corresponding access rights and control rights;

S433, low-level risk data: Use a pure six-digit number as the password to encrypt the data, and set corresponding access rights;

S434. No encryption processing is performed, the access permission is opened, but the access records are backed up and recorded. Based on the risk level determination results, customized encryption processing is performed on the risk data of the corresponding level. Core high-level risk data uses stronger encryption methods, such as three-letter plus three-digit passwords, and sets strict access and control permissions, while conducting authentication and recording. Medium-level risk data is encrypted using a six-digit pure alphabetic password, and low-level risk data is encrypted using a pure six-digit numeric password. For public-level data, encryption is not performed, but access records will be backed up.

In this embodiment, through customized encryption processing for different levels of risks, the level of data security and privacy protection can be improved. Strong encryption methods for core high-level risk data can effectively prevent unauthorized access and data leakage, and protect the security of sensitive information. Encryption of medium-level and low-level risk data provides moderate data protection, ensuring that data is not vulnerable to malicious attacks during transmission and storage. Backing up access records can help monitor and track data access behavior and improve data traceability and credibility. The access rights and control rights settings in the encryption processing method make the access and control of different levels of risk data more flexible. Corresponding permission levels can be set according to different business needs and security policies to ensure that only authorized personnel can access and operate the corresponding level of risk data, improving the compliance and security of data access. By customizing encryption of risk data based on risk level determination results, data security, privacy protection and access control capabilities can be improved. This helps protect a business’s financial data from risks and threats and ensures the confidentiality and integrity of data during transmission, storage and processing.

Example 6

This embodiment is an explanation of what was done in Embodiment 1. Specifically, the S5 includes:

S51. Through learning and analyzing normal behavior patterns, establish behavioral pattern benchmarks for each level of risk data; behavioral pattern benchmarks include access frequency, operation type and data transmission mode;

S52. Based on the established behavioral pattern benchmark, use intelligent anomaly monitoring to conduct real-time monitoring of encrypted level risk data; detect abnormalities in access frequency, operation type, and data transmission, and determine it as a first-level risk behavior; through monitoring Abnormalities in access frequency, operation type and data transmission can promptly detect first-level risk behaviors and improve risk sensitivity and identification capabilities.

S53. Use network traffic sensors, log sensors, and authentication sensors to monitor access intrusions, including unauthorized access, abnormal login attempts, and data leaks. If access intrusions are identified, they are determined to be secondary risk behaviors; by identifying these intrusions, Able to detect and prevent potential security threats in a timely manner and protect the security of corporate financial data.

S54. Combine the first-level risk behavior and the second-level risk behavior to generate a corresponding risk report. When a risk report is identified, the risk report is analyzed and an early warning command notification is generated and sent to the security team.

The early warning command notification includes one or more of email, text message, and instant message.

In this embodiment, the enterprise financial data security management method based on artificial intelligence can monitor and identify risk behaviors, generate corresponding risk reports, and send early warning notifications in a timely manner to improve risk response capabilities and security protection levels.

Example 6

Please refer to Figure 2, an enterprise financial data security management system based on artificial intelligence, including a data collection module, a data preprocessing module, a feature extraction module, a model training and classification module, a risk level determination module, a backup module, an encryption module, and a security module. Control module, abnormality monitoring module and early warning module;

The data collection module collects corporate financial data from different data sources and obtains a collection of corporate financial data samples;

The data preprocessing module is used to clean the collection samples of corporate financial data, remove duplicate values and handle missing values;

The feature extraction module is used to extract key features from financial data, including numerical features and text type features;

The model training and classification module is used to establish a deep learning model or other machine learning model, input the processed characteristic financial data into the model, and calculate and obtain the comprehensive classification coefficient Zhxs;

The risk level determination module is used to compare the comprehensive classification coefficient Zhxs with the preset standard threshold to determine the risk level of financial data. According to the set threshold conditions, the data is divided into different levels of risks, including core high-level risks and medium-level risks. Risk, low-level risk and public-level data;

The backup module is used for periodic backup of data corresponding to core high-level risks and medium-level risks;

The encryption module is used to encrypt risk data of the corresponding level;

The security control module is used to set corresponding access rights and control measures to ensure that only authorized personnel can access and process sensitive data;

The anomaly monitoring module is used to monitor the behavior patterns and access activities of encrypted risk data, identify abnormal behaviors and intrusion behaviors, and generate corresponding reports, including details of abnormal behaviors and risk assessment results;

The early warning module is used to send early warning notifications to relevant responsible persons or security teams based on corresponding reports, so that timely measures can be taken to deal with risk events.

In this embodiment, through the synergy of the above modules, the enterprise financial data security management method based on artificial intelligence can provide effective risk management and protection measures, reduce and ensure the security, integrity and availability of financial data. Based on artificial intelligence technology, improve Intelligent, reduce the problem of false positives and negatives caused by abnormal monitoring and risk assessment caused by administrators' manual authorization to manage financial data in the system.

Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art will understand that various changes, modifications, and substitutions can be made to these embodiments without departing from the principles and spirit of the invention. and modifications, the scope of the invention is defined by the appended claims and their equivalents.

Claims (10)

1. An artificial intelligence-based enterprise financial data security management method, characterized by: including the following steps, S1. Collect and obtain collective samples of corporate financial data, and process the collective samples of corporate financial data; S2. Extract features from the collected samples of processed financial data to obtain characteristic financial data; S3. Establish a deep learning model, classify the processed characteristic financial data, conduct model training, and obtain the comprehensive classification coefficient Zhxs; S4. Intelligent security encryption management compares the comprehensive classification coefficient Zhxs with the standard threshold. If it exceeds the standard threshold, it is determined to be core high-level risk data; if the comprehensive classification coefficient Zhxs is lower than 30% of the standard threshold, it is determined to be medium-level risk data; If the comprehensive classification coefficient Zhxs is lower than 40% of the standard threshold, it is determined to be low-level risk data; if the comprehensive classification coefficient Zhxs is lower than 50% of the standard threshold, it is determined to be public-level data; and the corresponding level risk data is corresponding encryption processing; S5. Analyze the behavior patterns of the corresponding level of risk data for encryption processing, establish benchmarks, monitor abnormal activities and access intrusion behaviors, identify risk behaviors through artificial intelligence technology, generate corresponding reports, and send early warning notifications. 2. An artificial intelligence-based enterprise financial data security management method according to claim 1, characterized in that: the step S1 includes: S11. Collect a collection of corporate financial data samples, including classified revenue, expenditure and profit data; it also includes unclassified data, which requires manual labeling of categories; Revenue includes the company's sales revenue, service fees and interest income; expenses include the company's purchase costs, monthly salary statements, rent and transportation costs; profits include the company's net profit and gross profit; unclassified data includes liabilities and shareholders' equity. , taxation and financial statement classification; liabilities include the company's short-term liabilities, long-term liabilities and bonds; shareholders' equity includes the company's equity information, capital reserves and surplus reserves; taxation includes the company's taxes, tax declarations and tax adjustments; financial statements include The company's balance sheet, income statement and cash flow statement; S12. Process the collected samples of corporate financial data. The processing includes data cleaning, missing value processing and normalization processing. 3. An artificial intelligence-based enterprise financial data security management method according to claim 1, characterized in that: the step S2 includes extracting features from a collection of financial data samples, including numerical feature extraction and text type features. Extraction, the numerical feature extraction includes extracting the total amount, average value and variance, the text type feature extraction includes extracting keywords and word frequency information to obtain characteristic financial data; Total extraction is used to calculate the total of financial data, including calculating total revenue, total expenses, and total profit; Average extraction is used to calculate the average of financial data, including calculating average sales, evaluation costs, and average profits; Variance extraction is used to calculate the variance of financial data, measure the degree of change in the data, and obtain a variance change waveform table; Text type feature extraction is used to process text fields in financial data, segment the text, extract stems, and convert the text into a keyword list. The keywords are set to product names, supplier names, and customer names; word frequency information extraction includes Count the number of times keywords appear in text in financial data, and calculate the word frequency of each keyword in the financial data sample collection as a text type feature. 4. An artificial intelligence-based enterprise financial data security management method according to claim 1, characterized in that: the step S3 includes: S31. Extract the total amount, average value and variance as numerical features, and extract keywords and word frequency information as text type features; associate these features with the classification labels of the financial data samples; S3. Establish a deep learning model for classification task training, including one of decision trees, support vector machines and neural networks; S4. Predict new characteristic financial data samples based on the deep learning model and obtain the classification results; calculate the comprehensive classification coefficient Zhxs based on the classification results. 5. An artificial intelligence-based enterprise financial data security management method according to claim 4, characterized in that: the comprehensive classification coefficient Zhxs is obtained by the following formula: In the formula, ,/> ,/> ,...,/> It is a feature value extracted from feature financial data, set to be a numerical feature or text feature;/> ,/> ,/> ,...,/> is the weight of the corresponding feature;/> for the correction constant. 6. An artificial intelligence-based enterprise financial data security management method according to claim 1, characterized in that: the S4 includes: S41. Set standard thresholds for judging different levels of risks; set standard thresholds to 30%, 40% and 50%; S42. Compare the comprehensive classification coefficient Zhxs of each characteristic financial data sample, and judge the risk level according to different conditions: If Zhxs exceeds the standard threshold, it is determined to be core high-level risk data; If Zhxs is 30% lower than the standard threshold, it is determined to be medium-level risk data; If Zhxs is lower than 40% of the standard threshold, it is determined to be low-level risk data; If Zhxs is lower than 50% of the standard threshold, it is determined to be public-level data, which means risk-free data; S43. According to the determination result of the risk level, perform corresponding encryption method processing on the risk data of the corresponding level. 7. An artificial intelligence-based enterprise financial data security management method according to claim 6, characterized in that: the encryption method of S43 is determined according to different levels of risk data; specifically, the encryption processing method includes: S431. Core high-level risk data: Use three letters plus three digits as the password to encrypt the data, set corresponding access rights and control rights, and perform identity verification and recording based on the access timestamp; S432, medium-level risk data: Use six pure letters as the password to encrypt the data, and set corresponding access rights and control rights; S433, low-level risk data: Use a pure six-digit number as the password to encrypt the data, and set corresponding access rights; S434. No encryption processing is performed, the access permission is opened, but the access records are backed up and recorded. 8. An artificial intelligence-based enterprise financial data security management method according to claim 1, characterized in that: the S5 includes: S51. Through learning and analyzing normal behavior patterns, establish behavioral pattern benchmarks for each level of risk data; behavioral pattern benchmarks include access frequency, operation type and data transmission mode; S52. Based on the established behavioral pattern benchmark, use intelligent anomaly monitoring to conduct real-time monitoring of the encrypted level risk data; if abnormalities in access frequency, operation type, and data transmission are detected, it will be determined as a first-level risk behavior; S53. Use network traffic sensors, log sensors, and authentication sensors to monitor access intrusions, including unauthorized access, abnormal login attempts, and data leaks. If access intrusions are identified, they are determined to be secondary risk behaviors; S54. Combine the first-level risk behavior and the second-level risk behavior to generate a corresponding risk report. When a risk report is identified, the risk report is analyzed and an early warning command notification is generated and sent to the security team. 9. An artificial intelligence-based enterprise financial data security management method according to claim 8, characterized in that: the early warning command notification includes one or more of email, text message and instant message. 10. An enterprise financial data security management system based on artificial intelligence, which is characterized by: including a data collection module, a data preprocessing module, a feature extraction module, a model training and classification module, a risk level determination module, a backup module, and an encryption module. Safety control module, abnormality monitoring module and early warning module; The data collection module collects corporate financial data from different data sources and obtains a collection of corporate financial data samples; The data preprocessing module is used to clean the collection samples of corporate financial data, remove duplicate values and handle missing values; The feature extraction module is used to extract key features from financial data, including numerical features and text type features; The model training and classification module is used to establish a deep learning model or other machine learning model, input the processed characteristic financial data into the model, and calculate and obtain the comprehensive classification coefficient Zhxs; The risk level determination module is used to compare the comprehensive classification coefficient Zhxs with the preset standard threshold to determine the risk level of financial data. According to the set threshold conditions, the data is divided into different levels of risks, including core high-level risks and medium-level risks. Risk, low-level risk and public-level data; The backup module is used for periodic backup of data corresponding to core high-level risks and medium-level risks; The encryption module is used to encrypt risk data of the corresponding level; The security control module is used to set corresponding access rights and control measures to ensure that only authorized personnel can access and process sensitive data; The anomaly monitoring module is used to monitor the behavior patterns and access activities of encrypted risk data, identify abnormal behaviors and intrusion behaviors, and generate corresponding reports, including details of abnormal behaviors and risk assessment results; The early warning module is used to send early warning notifications to relevant responsible persons or security teams based on corresponding reports, so that timely measures can be taken to deal with risk events.