Financial management electronic bill classified storage method and system

Abstract

The invention discloses a classified storage method and system for financial management electronic bills. The method comprises the following steps: acquiring text content and structured data of the electronic bills, extracting key information of the text content, combining the key information with the structured data, and constructing a multi-dimensional feature vector; constructing an electronic bill classification model, and training the electronic bill classification model based on the multi-dimensional feature vector; dynamically updating the trained electronic bill classification model based on the newly added bill type; classifying the electronic bills based on the dynamically updated electronic bill classification model, and adjusting a classification result; constructing a mapping relationship between the bill category and the business process, and automatically triggering the corresponding business process based on the classification result of the electronic bill; and analyzing a classification result of the electronic bills, and displaying an analysis result based on an interactive chart. According to the invention, refined intelligent classification of the electronic bills is realized, and the automation level and decision-making efficiency of financial management are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of financial management, and in particular relates to a method and system for classifying and storing electronic bills for financial management. Background Art

[0002] In the process of classifying and preserving electronic bills in financial management, how to achieve efficient, detailed, and personalized classification methods is a key technical issue. Traditional classification methods usually rely on pre-set fixed rules, which are difficult to adapt to the characteristics of large numbers, multiple types, and complex information of electronic bills. At the same time, different financial management scenarios have different requirements for bill classification. How to meet personalized classification needs while taking into account universality is a difficult problem that needs to be solved urgently.

[0003] In addition, how to organically combine the classified information of electronic invoices with functions such as authority management and data analysis is also an issue worthy of in-depth discussion. If the classified information is out of touch with authority management, it may lead to data leakage or confusion of authority; if the classified information and data analysis cannot match, it will affect the decision-making efficiency and accuracy of financial management.

[0004] Therefore, how to design a flexible, efficient and secure method for classifying and preserving electronic invoices is a key technical issue that needs to be solved urgently, which is of great significance for improving the level of refinement of financial management and personalized service capabilities. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a method and system for classifying and saving electronic invoices for financial management. The method can dynamically adapt to multi-dimensional classification needs and is deeply integrated with functions such as authority management and data analysis to solve the problems existing in the above-mentioned prior art.

[0006] To achieve the above object, the present invention provides a method for classifying and storing electronic bills for financial management, comprising the following steps:

[0007] Acquire the text content and structured data of the electronic invoice, extract key information of the text content, combine the key information with the structured data, and construct a multi-dimensional feature vector;

[0008] Building an electronic bill classification model based on a support vector machine, and training the electronic bill classification model based on the multi-dimensional feature vector;

[0009] Dynamically update the trained electronic bill classification model based on financial management scenarios and newly added bill types;

[0010] Classify the electronic invoices to be processed based on the dynamically updated electronic invoice classification model, and adjust the output method of the classification results based on the user role and permission settings;

[0011] Build a mapping relationship between bill categories and business processes, and automatically trigger corresponding business processes based on the classification results of electronic bills;

[0012] Perform data analysis on the classification results of electronic invoices and display the analysis results based on interactive charts.

[0013] Optionally, the process of obtaining text content and structured data of the electronic receipt, extracting key information of the text content, combining the key information with the structured data, and constructing a multi-dimensional feature vector includes:

[0014] The text content and structured data of the electronic invoice are obtained, natural language processing is performed on the text content, and the key field information of the invoice is extracted using a named entity recognition algorithm; based on the extracted key field information and combined with the structured data of the electronic invoice, a multi-dimensional feature vector is constructed, and the feature vector includes a bill type feature, an amount feature, a date feature, and an invoice number feature.

[0015] Optionally, the process of training the electronic bill classification model based on the multi-dimensional feature vector includes:

[0016] For the constructed multi-dimensional feature vector, the importance of each feature to the classification result is calculated, and the features are sorted from high to low according to the importance evaluation results to obtain a feature importance sorting list; a feature selection threshold is preset, and according to the feature importance sorting list, features with importance higher than the feature selection threshold are selected to form a key feature subset; a cross-validation method is adopted to train the electronic invoice classification model using the key feature subset.

[0017] Optionally, the process of dynamically updating the trained electronic bill classification model based on the financial management scenario and the newly added bill type includes:

[0018] Extract feature information of newly added bills, input the feature information of the newly added bills into the electronic bill classification model, calculate the classification error rate of the newly added bills, and if the classification error rate is higher than a preset error threshold, use an online learning algorithm to update the parameters of the electronic bill classification model using the newly added bills.

[0019] Optionally, based on user roles and permission settings, the process of adjusting the output mode of the classification results includes:

[0020] According to the pre-configured user role and permission information, the role type and operation permission list of the current user are obtained, and the operation permission list of the current user is matched with the category information of the electronic bill classification model to determine the user's access control permissions for each bill category; when generating the classification result, according to the user's access control permissions, the bill categories that the user does not have access to are hidden or desensitized in the classification result, and the adjusted classification result is returned to the user, showing the bill categories and related information that the user has access to.

[0021] Optionally, a mapping relationship between bill categories and business processes is constructed, and a process of automatically triggering corresponding business processes based on the classification results of electronic bills includes:

[0022] Obtain the classification results of the electronic invoice, and determine the approval, accounting and reimbursement processes triggered by the invoice according to the mapping relationship between the preset invoice category and the business process; call the workflow engine to instantiate the determined approval, accounting and reimbursement processes and generate corresponding process tasks; obtain relevant business data for the generated process tasks and pass the relevant business data to the workflow engine; drive the execution of process tasks through the workflow engine to automatically complete the processing of approval, accounting and reimbursement links; after the process execution is completed, synchronize the processing results of the invoice with the financial system.

[0023] Optionally, data analysis is performed on the classification results of the electronic invoices, and a process of displaying the analysis results based on an interactive chart includes:

[0024] Based on the classification results of electronic bills, multi-dimensional data analysis technology is used to analyze the data characteristics of bill type distribution, amount statistics and time series; through data visualization technology, the multi-dimensional analysis results are converted into intuitive statistical reports and charts to generate bill classification statistical reports; data mining algorithms are used to conduct in-depth analysis of the classification results of electronic bills, explore hidden patterns and trends in bill data, and generate insight analysis reports; bill classification statistical reports and insight analysis reports are displayed in the form of interactive charts.

[0025] The present invention also provides a financial management electronic bill classification and storage system for implementing a financial management electronic bill classification and storage method, comprising: a feature vector construction module, a classification model training module, a classification model update module, a classification result adjustment module, a business process triggering module and a classification result display module;

[0026] The feature vector construction module is used to obtain the text content and structured data of the electronic bill, extract key information of the text content, combine the key information with the structured data, and construct a multi-dimensional feature vector;

[0027] The classification model training module is used to construct an electronic bill classification model based on a support vector machine, and to train the electronic bill classification model based on the multi-dimensional feature vector;

[0028] The classification model updating module is used to dynamically update the trained electronic bill classification model based on financial management scenarios and newly added bill types;

[0029] The classification result adjustment module is used to classify the electronic bills to be processed based on the dynamically updated electronic bill classification model, and adjust the output mode of the classification results based on the user role and permission settings;

[0030] The business process triggering module is used to construct a mapping relationship between bill categories and business processes, and automatically trigger corresponding business processes based on the classification results of electronic bills;

[0031] The classification result display module is used to perform data analysis on the classification results of the electronic invoice and display the analysis results based on interactive charts.

[0032] The present invention also provides an electronic device, comprising: a memory and a processor; the memory is used to store programs; the processor is used to execute the programs to implement various steps of the financial management electronic bill classification and storage method.

[0033] The present invention also provides a readable storage medium on which a computer program is stored. When the computer program is executed by a processor, each step of the financial management electronic bill classification and storage method is implemented.

[0034] Compared with the prior art, the present invention has the following advantages and technical effects:

[0035] The present invention discloses a classification and preservation method and system for electronic bills for financial management. The method performs natural language processing on bill text, extracts key information and constructs a multidimensional feature vector as model input; dynamically updates the model using incremental learning to maintain the timeliness of classification rules; dynamically adjusts the display of classification results according to user permissions to ensure data security; establishes a mapping relationship between bill categories and business processes to achieve classification-driven process automation; performs multidimensional data analysis and visualization based on classification results to provide support for financial decision-making. The present invention realizes refined intelligent classification of electronic bills and improves the automation level and decision-making efficiency of financial management. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0037] Figure 1A schematic diagram of a flow chart of a method for classifying and saving electronic bills for financial management according to an embodiment of the present invention;

[0038] Figure 2 A schematic diagram of a process for updating an electronic bill classification model according to an embodiment of the present invention;

[0039] Figure 3 The figure is a flow chart of a method for adjusting the output of classification results according to an embodiment of the present invention. DETAILED DESCRIPTION

[0040] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0041] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0042] Embodiment 1

[0043] like Figure 1 As shown, this embodiment provides a method for classifying and saving electronic bills for financial management, comprising the following steps:

[0044] Obtain the text content and structured data of the electronic invoice, extract the key information of the text content, combine the key information with the structured data, and construct a multi-dimensional feature vector;

[0045] Construct an electronic bill classification model based on support vector machine, and train the electronic bill classification model based on multi-dimensional feature vectors;

[0046] Dynamically update the trained electronic bill classification model based on financial management scenarios and newly added bill types;

[0047] Classify the electronic invoices to be processed based on the dynamically updated electronic invoice classification model, and adjust the output method of the classification results based on the user role and permission settings;

[0048] Build a mapping relationship between bill categories and business processes, and automatically trigger corresponding business processes based on the classification results of electronic bills;

[0049] Perform data analysis on the classification results of electronic invoices and display the analysis results based on interactive charts.

[0050] The process of obtaining the text content and structured data of the electronic invoice, extracting the key information of the text content, combining the key information with the structured data, and constructing a multi-dimensional feature vector includes:

[0051] The text content and structured data of the electronic invoice are obtained, natural language processing is performed on the text content, and the named entity recognition algorithm is used to extract key field information such as the invoice type, amount, date, and invoice number. Based on the extracted key field information and combined with the structured data of the electronic invoice, a multi-dimensional feature vector is constructed, which includes the invoice type feature, amount feature, date feature, invoice number feature, etc.

[0052] Specifically, first, the system obtains the text content and structured data of the electronic invoice. For the text content, the named entity recognition algorithm based on BiLSTM-CRF is used to extract key field information such as bill type, amount, date, invoice number, etc. by training a labeled data set containing 10,000 electronic invoices. The recognition accuracy of bill type reaches 95%, the average absolute error of amount recognition is 3%, the accuracy of date recognition is 91%, and the recognition accuracy of invoice number is 98%. Combining the extracted key field information and the structured data of the electronic invoice, a feature vector containing 150 dimensions is constructed, covering feature representations of multiple aspects such as bill type, amount, date, invoice number, etc. For the constructed feature vector, the CART decision tree algorithm is used to evaluate the feature importance, and 30 key features with a greater impact on the classification results are screened out through indicators such as the Gini index and information gain ratio. The screened key features are input into the SVM classification model, and the RBF kernel function is used to train on a data set containing 5,000 labeled electronic invoices through 5-fold cross validation. The accuracy of the obtained classification model on the test set is 97%. When a new electronic invoice needs to be classified, the system automatically extracts its text content and structured data, constructs a feature vector, and inputs it into the trained classification model for prediction. For example, for a special VAT invoice, the classification model calculates the distance from the sample point to the hyperplane based on its feature representation and determines that the probability of it being a VAT invoice is 95%, which is higher than other types, so it is classified as a VAT invoice. Based on the classification results, the system automatically classifies the electronic invoice into the corresponding category folder to facilitate subsequent query and management. The entire process does not require manual intervention, and the efficiency of invoice processing has increased by 80%.

[0053] The feasible process of training the electronic bill classification model based on the multi-dimensional feature vector includes:

[0054] For the constructed multi-dimensional feature vector, the decision tree algorithm is used for feature selection to screen out the key feature subsets that have a greater impact on the classification results. The screened key feature subsets are input into the support vector machine classification model for training to obtain the trained electronic bill classification model. The electronic bill to be classified is obtained, its text content and structured data are extracted, a multi-dimensional feature vector is constructed, and the feature vector is input into the trained electronic bill classification model.

[0055] Furthermore, for the constructed multi-dimensional feature vector, a decision tree algorithm is used to select features, and the process of screening out a subset of key features that have a greater impact on the classification results includes:

[0056] For the constructed multi-dimensional feature vector, the decision tree algorithm is used to evaluate the feature importance and calculate the influence of each feature on the classification result. According to the feature importance evaluation results, the features are sorted in order from high to low in order of feature importance to obtain a feature importance sorting list. The feature selection threshold is set, and according to the feature importance sorting list, the features with importance higher than the threshold are selected to form a key feature subset. The cross-validation method is used to train the classification model using the key feature subset, and the model performance is evaluated on the validation set to obtain the model performance evaluation result. If the model performance evaluation result meets the preset performance requirements, the key feature subset is used as the final feature selection result; if not, the feature selection threshold is adjusted and the above steps are repeated until a feature subset that meets the performance requirements is obtained.

[0057] The process of dynamically updating the trained electronic bill classification model based on financial management scenarios and newly added bill types can be implemented as follows:

[0058] Obtain information about changes in financial management scenarios and new bill types, and extract feature information of bills for new bill sample data. Input feature information of new bill samples into the bill classification model to determine whether the bill classification model needs to be updated. If the bill classification model needs to be updated, use the online learning algorithm to update the model parameters using the new bill sample data. According to the importance of the bill samples, use a weighted adjustment method to dynamically adjust the learning rate of the online learning algorithm. Through incremental learning, continuously update the bill classification model to adapt to changes in financial management scenarios and new bill types, and maintain the flexibility and timeliness of classification rules.

[0059] Furthermore, the feature information of the new bill sample is input into the electronic bill classification model, and the process of determining whether the electronic bill classification model needs to be updated includes:

[0060] like Figure 2As shown in the figure, obtain the feature information of the new bill sample, including the bill field, layout structure, anti-counterfeiting mark, etc., and convert it into a data format acceptable to the model. Input the preprocessed new bill feature data into the trained bill classification model to obtain the predicted classification result of the bill. Obtain the actual classification label of the new bill sample, compare it with the model prediction result, and calculate the difference between the two. If the difference exceeds the preset threshold, it is determined that the existing model cannot accurately identify the new bill and needs to be updated; otherwise, it is considered that the model is still applicable and does not need to be updated. For situations that require updating, collect more sample data of this type of new bill, annotate it, and merge it into the original training data set. Use algorithms such as gradient boosting decision tree and support vector machine to retrain the bill classification model, and evaluate the model performance through methods such as cross-validation. Deploy the optimized new model to the production environment, identify and classify new bills, and continuously monitor its operation to detect and solve problems in a timely manner.

[0061] Furthermore, according to the importance of the bill samples, a weighted adjustment method is adopted to dynamically adjust the learning rate of the online learning algorithm, including:

[0062] Obtain the bill samples in the training data set, calculate the importance score of each sample according to the preset rules, and obtain the sample importance weight vector. Input the sample importance weight vector into the online learning algorithm, weight the gradient contribution of each bill sample according to the weight vector, and adjust the update amplitude of the model parameters. In each iteration cycle, dynamically adjust the learning rate according to the importance of the current bill sample. If the sample importance is higher than the threshold, increase the learning rate; if it is lower than the threshold, reduce the learning rate. Through weighted adjustment and dynamic learning rate, the model pays more attention to important bill samples when updating parameters, accelerates convergence, and improves training efficiency. During the training process, regularly evaluate the performance indicators of the model on the validation set, such as accuracy and recall. If the performance improvement is not obvious, further adjust the sample weight and learning rate. After the training is completed, use the test set data to evaluate the model and obtain the generalization performance indicators of the model. Compare the indicators with the baseline model to determine the optimization effect of weighted adjustment and dynamic learning rate. If the model performance meets expectations, the trained model is applied to actual business scenarios. Continuously monitor model performance, iteratively optimize the model based on business feedback data, and improve the robustness and adaptability of the model.

[0063] The process of adjusting the output of classification results based on user roles and permissions can be implemented as follows:

[0064] like Figure 3As shown, according to the pre-configured user role and permission information, the role type and operation permission list of the current user are obtained. The user permission list is matched with the category information of the bill classification model to determine the user's access control permissions for each bill category. When the bill classification model generates classification results, the display content of the classification results is dynamically adjusted according to the user's access control permissions. For bill categories that the user has no access to, they are hidden or desensitized in the classification results to ensure data security and privacy. The adjusted classification results are returned to the user, showing the bill categories and related information that the user has access to. Through the real-time update of the access control list, the user's operation permissions for the bill category are dynamically adjusted to ensure the timeliness and accuracy of the permission setting. The user permission verification and classification result adjustment functions are integrated with the bill classification model to form a complete bill classification and permission management solution.

[0065] Specifically, during the system initialization phase, user roles and permission information are defined through configuration files. For example, the administrator role has read and write permissions for all bill categories, the financial role has read and write permissions for invoices and expense reports, and the sales role has read permissions for orders and contracts. When a user logs in to the system, the role-based access control (RBAC) model is used to obtain the user's role type and operation permission list, and match it with the category information of the bill classification model. For example, the BitSet data structure is used to represent user permissions and bill categories, and the user's access control permissions for each bill category are quickly determined through bit operations, with a time complexity of O(1). When the bill classification model generates classification results, the data desensitization algorithm is used to process the bill categories that are not accessible according to the user's access control permissions, such as masking and encryption. For example, for the amount field, the regular expression "\d+" can be used to match the numeric part, and then replaced with "*" to hide the amount. For sensitive information, such as ID card number, bank card number, etc., the SHA-256 algorithm can be used for irreversible encryption to ensure data security and privacy. When permissions are changed, the user's operating permissions for the ticket category are dynamically adjusted through incremental updates, the change information is compared with the user permission list, the newly added, deleted, and modified permission items are identified, and then the BitSet is updated to ensure the timeliness and accuracy of permission settings. By integrating the access control interceptor with the ticket classification model, the user permissions are verified and the results are adjusted before the classification results are returned, forming a complete ticket classification and permission management solution. For example, using Spring AOP to implement access control interceptors, permission verification and result processing are performed before and after method calls, and flexible permission control is achieved through annotation configuration to improve the security and maintainability of the system.

[0066] The process of constructing a mapping relationship between bill categories and business processes and automatically triggering corresponding business processes based on the classification results of electronic bills includes:

[0067] Obtain the classification information of the electronic invoice, and determine the approval, accounting, and reimbursement processes triggered by the invoice according to the mapping relationship between the preset invoice category and the business process; call the workflow engine to instantiate the determined approval, accounting, and reimbursement processes and generate corresponding process tasks; obtain relevant business data, such as the invoice amount, date, submitter, etc. for the generated process tasks, and pass these data to the workflow engine; drive the execution of process tasks through the workflow engine, and automatically complete the processing of approval, accounting, reimbursement, etc. according to the preset process definition and business rules; during the process execution, use machine learning algorithms to learn and optimize the classification information of the invoice, and continuously improve the accuracy of invoice classification through training on historical data; if an abnormal situation occurs during the process execution, such as incomplete invoice information, unavailable approver, etc., the preset exception handling mechanism is triggered, and the process is handled by supplementing data or transferring tasks; after the process execution is completed, the processing results of the invoice are synchronized with the financial system to ensure the consistency of financial data and business processes, and generate corresponding financial statements and analysis data to provide data support for financial management.

[0068] Specifically, after obtaining the classification information of the electronic invoice, the system will match the invoice category with the corresponding business process according to the preset mapping relationship. For example, the decision tree algorithm is used to judge the invoice category. If the invoice category is "travel expense reimbursement", the corresponding approval, accounting and reimbursement process is triggered. Then, the system calls the workflow engine to generate a process task instance according to the determined process definition and obtain the relevant business data of the invoice. The workflow engine automatically drives the execution of the task according to the preset process definition and business rules. During the process execution, the system uses the support vector machine (SVM) algorithm to learn and optimize the invoice classification information, and continuously improves the classification accuracy by training 10,000 historical data. If an abnormal situation occurs during the process execution, such as the invoice amount is empty or the approver resigns, the system will trigger the preset exception handling mechanism, and handle it by data supplementation or task transfer to ensure the smooth progress of the process. After the process execution is completed, the system automatically synchronizes the processing results of the invoice with the financial system, and generates corresponding financial statements and analysis data to provide data support and decision-making basis for financial management.

[0069] The process of performing data analysis on the classification results of electronic invoices and displaying the analysis results based on interactive charts includes:

[0070] Obtain electronic bill data, classify electronic bills, and obtain bill classification result data. Based on the bill classification result data, use multi-dimensional data analysis technology to analyze the data characteristics of bill type distribution, amount statistics, and time series. Use data visualization technology to convert multi-dimensional analysis results into intuitive statistical reports and charts to generate bill classification statistical reports. Use data mining algorithms to conduct in-depth analysis of bill classification result data, mine hidden patterns and trends in bill data, and generate insight analysis reports. Display bill classification statistical reports and insight analysis reports in the form of interactive charts to support users to interactively explore and analyze data. Dynamically update statistical reports and insight analysis results based on user operations and selections in interactive charts to provide real-time data analysis support. Provide bill classification statistical reports and insight analysis results to the finance department as data support for financial decision-making, helping the finance department optimize bill management processes and improve financial management efficiency.

[0071] Specifically, after the electronic invoice data is acquired, the electronic invoices are first classified and processed through machine learning algorithms such as support vector machines (SVM), for example, the invoices are classified into categories such as transportation, catering, and accommodation, and the classification accuracy is recorded to be above 95%. Then, the classified invoice data is analyzed in depth using multi-dimensional data analysis technology, such as using the principal component analysis (PCA) method to analyze the distribution characteristics of the invoice amount, and it is found that large invoices are mainly concentrated at the end of the month. Through data visualization tools such as Tableau, the analysis results are converted into bar charts and line charts to intuitively display the trend of invoice types and amounts over time. Furthermore, association rule mining algorithms such as the Apriori algorithm are used to mine the consumption patterns in the invoice data, and it is found that the correlation between transportation and catering invoices on weekdays is as high as 80%. These analysis results are displayed through interactive charts, which support users to dynamically observe the changes in consumption patterns by filtering different invoice categories and time periods. In addition, according to the real-time operation of users, the system can instantly update statistical reports and analysis results to ensure that the financial department can make decisions based on the latest data. Ultimately, these statistical reports and insight analysis results are integrated and provided to the finance department to help them optimize the bill management process, such as adjusting budget allocations, and improving the efficiency and accuracy of overall financial management.

[0072] The present invention also provides a financial management electronic bill classification and storage system for implementing a financial management electronic bill classification and storage method, comprising: a feature vector construction module, a classification model training module, a classification model update module, a classification result adjustment module, a business process triggering module and a classification result display module;

[0073] The feature vector construction module is used to obtain the text content and structured data of the electronic invoice, extract the key information of the text content, combine the key information with the structured data, and construct a multi-dimensional feature vector;

[0074] The classification model training module is used to build an electronic bill classification model based on a support vector machine and train the electronic bill classification model based on a multi-dimensional feature vector;

[0075] The classification model update module is used to dynamically update the trained electronic bill classification model based on financial management scenarios and newly added bill types;

[0076] The classification result adjustment module is used to classify the electronic invoices to be processed based on the dynamically updated electronic invoice classification model, and adjust the output mode of the classification results based on the user role and permission settings;

[0077] The business process trigger module is used to build a mapping relationship between bill categories and business processes, and automatically trigger the corresponding business process based on the classification results of electronic bills;

[0078] The classification result display module is used to perform data analysis on the classification results of electronic invoices and display the analysis results based on interactive charts.

[0079] Embodiment 2

[0080] The present invention also provides an electronic device, comprising: a memory and a processor; the memory is used to store programs; the processor is used to execute the programs to implement various steps of the financial management electronic bill classification and storage method.

[0081] Embodiment 3

[0082] The present invention also provides a readable storage medium on which a computer program is stored. When the computer program is executed by a processor, each step of the financial management electronic bill classification and storage method is implemented.

[0083] The above are only preferred specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for classifying and storing electronic bills for financial management, characterized in that: The following steps are involved: Acquire the text content and structured data of the electronic invoice, extract key information of the text content, combine the key information with the structured data, and construct a multi-dimensional feature vector; Building an electronic bill classification model based on a support vector machine, and training the electronic bill classification model based on the multi-dimensional feature vector; Dynamically update the trained electronic bill classification model based on financial management scenarios and newly added bill types; Classify the electronic invoices to be processed based on the dynamically updated electronic invoice classification model, and adjust the output method of the classification results based on the user role and permission settings; Build a mapping relationship between bill categories and business processes, and automatically trigger corresponding business processes based on the classification results of electronic bills; Perform data analysis on the classification results of electronic invoices and display the analysis results based on interactive charts.

2. The method for classifying and storing electronic bills for financial management according to claim 1, characterized in that: The process of obtaining text content and structured data of an electronic receipt, extracting key information of the text content, combining the key information with the structured data, and constructing a multi-dimensional feature vector includes: The text content and structured data of the electronic invoice are obtained, natural language processing is performed on the text content, and the key field information of the invoice is extracted using a named entity recognition algorithm; based on the extracted key field information and combined with the structured data of the electronic invoice, a multi-dimensional feature vector is constructed, and the feature vector includes a bill type feature, an amount feature, a date feature, and an invoice number feature.

3. The method for classifying and storing electronic bills for financial management according to claim 2, characterized in that: The process of training the electronic bill classification model based on the multi-dimensional feature vector includes: For the constructed multi-dimensional feature vector, the importance of each feature to the classification result is calculated, and the features are sorted from high to low according to the importance evaluation results to obtain a feature importance sorting list; a feature selection threshold is preset, and according to the feature importance sorting list, features with importance higher than the feature selection threshold are selected to form a key feature subset; a cross-validation method is adopted to train the electronic invoice classification model using the key feature subset.

4. The method for classifying and storing electronic bills for financial management according to claim 1, characterized in that: The process of dynamically updating the trained electronic bill classification model based on financial management scenarios and newly added bill types includes: Extract feature information of newly added bills, input the feature information of the newly added bills into the electronic bill classification model, calculate the classification error rate of the newly added bills, and if the classification error rate is higher than a preset error threshold, use an online learning algorithm to update the parameters of the electronic bill classification model using the newly added bills.

5. The method for classifying and storing electronic bills for financial management according to claim 1, characterized in that: Based on the user role and permission settings, the process of adjusting the output of classification results includes: According to the pre-configured user role and permission information, the role type and operation permission list of the current user are obtained, and the operation permission list of the current user is matched with the category information of the electronic bill classification model to determine the user's access control permissions for each bill category; when generating the classification result, according to the user's access control permissions, the bill categories that the user does not have access to are hidden or desensitized in the classification result, and the adjusted classification result is returned to the user, showing the bill categories and related information that the user has access to.

6. The method for classifying and storing electronic bills for financial management according to claim 1, characterized in that: The process of building a mapping relationship between bill categories and business processes and automatically triggering corresponding business processes based on the classification results of electronic bills includes: Obtain the classification results of the electronic invoice, and determine the approval, accounting and reimbursement processes triggered by the invoice according to the mapping relationship between the preset invoice category and the business process; call the workflow engine to instantiate the determined approval, accounting and reimbursement processes and generate corresponding process tasks; obtain relevant business data for the generated process tasks and pass the relevant business data to the workflow engine; drive the execution of process tasks through the workflow engine to automatically complete the processing of approval, accounting and reimbursement links; after the process execution is completed, synchronize the processing results of the invoice with the financial system.

7. The method for classifying and storing electronic bills for financial management according to claim 1, characterized in that: The process of performing data analysis on the classification results of electronic invoices and displaying the analysis results based on interactive charts includes: Based on the classification results of electronic bills, multi-dimensional data analysis technology is used to analyze the data characteristics of bill type distribution, amount statistics and time series; through data visualization technology, the multi-dimensional analysis results are converted into intuitive statistical reports and charts to generate bill classification statistical reports; data mining algorithms are used to conduct in-depth analysis of the classification results of electronic bills, explore hidden patterns and trends in bill data, and generate insight analysis reports; bill classification statistical reports and insight analysis reports are displayed in the form of interactive charts.

8. A financial management electronic bill classification and storage system, characterized in that: The method for classifying and saving electronic invoices for financial management according to any one of claims 1 to 7 comprises: a feature vector construction module, a classification model training module, a classification model update module, a classification result adjustment module, a business process triggering module and a classification result display module; The feature vector construction module is used to obtain the text content and structured data of the electronic bill, extract key information of the text content, combine the key information with the structured data, and construct a multi-dimensional feature vector; The classification model training module is used to construct an electronic bill classification model based on a support vector machine, and to train the electronic bill classification model based on the multi-dimensional feature vector; The classification model updating module is used to dynamically update the trained electronic bill classification model based on financial management scenarios and newly added bill types; The classification result adjustment module is used to classify the electronic bills to be processed based on the dynamically updated electronic bill classification model, and adjust the output mode of the classification results based on the user role and permission settings; The business process triggering module is used to construct a mapping relationship between bill categories and business processes, and automatically trigger corresponding business processes based on the classification results of electronic bills; The classification result display module is used to perform data analysis on the classification results of the electronic invoice and display the analysis results based on interactive charts.

9. An electronic device, characterized in that: include: Memory and processor; The memory is used to store programs; the processor is used to execute the programs to implement the various steps of the financial management electronic invoice classification and storage method as described in any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, each step of the financial management electronic bill classification and storage method as described in any one of claims 1 to 7 is implemented.

Images