An agent-based reimbursement form filling method, apparatus, device and medium
By obtaining reimbursement requests through the intelligent agent assistant interface, using a large language model for intent recognition and process classification, and triggering workflow branch operations, the problem of manual entry and multi-level review in the enterprise reimbursement process is solved, and efficient and accurate reimbursement processing is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SI-TECH INFORMATION TECH CO LTD
- Filing Date
- 2025-12-31
- Publication Date
- 2026-05-29
AI Technical Summary
In the current corporate expense reimbursement process, manual data entry, pasting of receipts, and multi-level manual review result in long processing times, high costs, and a high risk of errors, affecting efficiency and accuracy.
An agent-based expense reimbursement form filling method is adopted. The reimbursement request is obtained through the agent assistant interface. A pre-trained large language model is used for intent recognition and process classification to trigger corresponding workflow branch operations, including generating a draft expense form, writing invoice information, and verifying expenses. Multiple interfaces are used to meet business needs.
It improves the efficiency and accuracy of the reimbursement process, reduces human error, enhances user experience, ensures data security and compliance, and saves employees time.
Smart Images
Figure CN122113872A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of expense reimbursement processing technology, and in particular to an expense reimbursement form filling method, apparatus, equipment and medium based on intelligent agents. Background Technology
[0002] In the field of corporate financial management, the expense reimbursement process is an important and routine task. As companies grow in size and their business expands, the volume of expense reimbursements also increases, making the efficiency and accuracy of the reimbursement process significantly impact the company's operating costs and management effectiveness.
[0003] In the existing corporate expense reimbursement process, during the document processing and pasting stage, employees are required to manually collect various paper invoices, electronic invoices, itineraries, etc., after business trips or other expenses, and paste them onto the reimbursement form according to regulations. During the manual data entry stage, employees need to manually fill in information such as the reimbursement date, amount, reason for receipt, and expense type in the reimbursement system. In the preliminary review stage, department and project personnel manually assess the reasonableness and compliance of the reimbursement form according to the company's financial regulations, while finance personnel verify the authenticity of the documents and the consistency of the amounts.
[0004] However, the manual data entry, pasting of receipts, and multi-level manual review process consume a significant amount of time and manpower, resulting in long reimbursement cycles and a poor employee experience. Furthermore, manual data entry is prone to errors, such as incorrect amounts or tax identification numbers, and manual review can be affected by factors such as fatigue and lack of experience, leading to misjudgments of compliance. Summary of the Invention
[0005] The technical problem to be solved by the present invention is to provide a method, apparatus, device and medium for filling out expense reimbursement forms based on intelligent agents, in order to solve at least one of the above-mentioned technical problems.
[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: Firstly, this application provides a reimbursement form filling method based on an intelligent agent, employing the following technical solution: A method for filling out expense reimbursement forms based on intelligent agents includes: Obtain the reimbursement request entered by the user through the intelligent agent assistant interface; The reimbursement request is input into a pre-trained large language model for intent recognition to obtain intent information of the reimbursement request. Based on the intent information, the process classification result of the reimbursement request is determined. The process classification result includes querying the business trip task book, generating a reimbursement form based on the business trip task book, and writing invoice information into the reimbursement form details. Based on the process classification result, the workflow branch corresponding to the process classification result is triggered to perform the corresponding operation, and the execution result of the reimbursement request is obtained. The execution result of the reimbursement request is fed back based on the intelligent agent assistant interface. The operation includes calling the database query interface, business plugin interface or external tool interface.
[0007] The beneficial effects of this invention are: accurately identifying and classifying the intent of user reimbursement requests, then triggering corresponding workflow branches to execute operations based on different classifications, utilizing multiple interfaces to meet different business needs, and providing feedback on execution results through the intelligent assistant interface, which can save employees' reimbursement form filling time, improve the efficiency of the reimbursement process, avoid errors and compliance risks from manual operations, and enhance the user experience.
[0008] Based on the above technical solution, the present invention can be further improved as follows.
[0009] Furthermore, when the process classification result is "generate expense report based on business trip assignment", the step of triggering the corresponding workflow branch to perform the corresponding operation based on the process classification result includes: Obtain the user's login information and business trip assignment number; Determine whether the user's login information matches the traveler's identity corresponding to the business trip assignment number; If the user's login information matches the traveler's identity corresponding to the business trip assignment number, then the business trip information is extracted from the travel application data corresponding to the business trip assignment. The business trip information includes the project, business trip time, and destination. Call the expense report draft generation interface, pass in the business trip information as a parameter, generate an expense report draft and store it in the database, and provide the expense report number corresponding to the expense report draft based on the intelligent agent assistant interface.
[0010] The beneficial effects of adopting the above-mentioned further solution are: ensuring that only the person on the business trip corresponding to the business trip assignment can generate the draft expense report, thus guaranteeing the data security and accuracy of the expense report process; automatically extracting business trip information from the travel application data corresponding to the business trip assignment and generating the draft expense report, simplifying the expense report filling process for employees, saving employees' time, and improving the efficiency of the expense report process; and storing the generated draft expense report in the database and providing the expense report number, making it convenient for employees to query and manage expense reports later.
[0011] Furthermore, when the process classification result indicates that invoice information is written into the expense report details, the step of triggering the corresponding workflow branch to perform the corresponding operation based on the process classification result includes: Receive invoice attachment information; Optical character recognition is performed on the invoice attachment information to obtain structured invoice information; The structured invoice information is subjected to compliance verification, which includes verification of duplicate invoice reimbursement, verification of consistency of the invoicing entity, and verification of the authenticity of the invoice. If the verification is successful, the structured invoice information will be filled into the corresponding expense details in the reimbursement form. If the verification fails, the result will be displayed on the AI assistant interface.
[0012] The beneficial effects of adopting the above-mentioned further solution are as follows: receiving invoice attachment information and performing optical character recognition can yield structured invoice information, which facilitates subsequent processing; verifying duplicate invoice reimbursement, consistency of invoicing entity, and authenticity of invoices on the structured invoice information can improve the accuracy and compliance of reimbursement and reduce compliance risks; after successful verification, the invoice information is filled into the expense details of the reimbursement form, which simplifies the reimbursement process; when verification fails, the result is fed back on the intelligent assistant interface, which can promptly remind the user.
[0013] Furthermore, after filling the structured invoice information into the corresponding expense details of the reimbursement form, the process also includes: Based on preset expense limit rules, it is determined whether the expense items in the expense details of the reimbursement form exceed the budget. The preset expense limit rules include limits on transportation expenses, accommodation expenses, and total project budget for different job levels.
[0014] If any expense item in the expense report exceeds the budget, an overspending reminder will be generated. The overspending reminder will distinguish between personal overspending and project budget overspending. An overspending template will be displayed on the intelligent assistant interface. After the user fills in the overspending template, the completed overspending template will be marked at the approval process node.
[0015] The beneficial effects of adopting the above-mentioned further solution are: after filling the structured invoice information into the expense details of the reimbursement form, based on the preset expense limit rules such as the transportation expense limit, accommodation expense limit and the total project budget limit, it can determine whether the expense items in the expense details of the reimbursement form are over-spending. This can help detect overspending in advance, avoid the overspending being returned for modification by the approver, and improve the efficiency of the reimbursement process.
[0016] Furthermore, it also includes: In response to a user's withdrawal request, withdraw the expense report submitted by the user before it enters the approval process stage; In response to a user's request for expedited review, an approval reminder notification is sent to the approver based on the intelligent assistant.
[0017] The beneficial effects of adopting the above-mentioned further solutions are: when a user finds an error after submitting an expense report, they can promptly withdraw and modify it before approval by submitting a withdrawal request, avoiding having to ask the approver to return it and improving modification efficiency; when the approver fails to approve the expense report in a timely manner, the user can send a reminder notification to the approver through a follow-up request, urging the approver to approve it in a timely manner and improving the efficiency of the expense report process.
[0018] Furthermore, it also includes: In response to booking events related to user travel in the business system, obtain the user's business trip assignment, historical expense data, travel application data, and travel consumption record data; A correlation analysis was conducted on business trip assignment documents, travel application data, and travel expense records to obtain the correlation analysis results; Based on the correlation analysis results, travel application data and travel consumption record data for the same travel event are correlated and integrated to obtain the integration result; Based on the integrated results, historical reimbursement data, and a pre-set recommendation model, predict the user's reimbursement items; Based on the user's reimbursement items and the integrated results, a draft reimbursement form is generated.
[0019] The beneficial effects of adopting the above-mentioned further solution are: based on the booking events related to user travel in the business system, relevant data is obtained, analyzed, and integrated; historical expense data and preset models are used to predict user expense items, thereby automatically generating a draft expense report, saving the applicant's time in filling out forms. Furthermore, it also includes: When a user is detected to have logged into the intelligent agent assistant, the draft expense report is recommended and displayed on the intelligent agent assistant interface. In response to the user's confirmation instruction for the draft expense report, the draft expense report is converted into an official expense report.
[0020] The beneficial effects of adopting the above-mentioned further solution are: it can proactively predict user expense reports and recommend and display a draft expense report when the user logs in to the intelligent assistant. After the user confirms, the draft is converted into a formal expense report, saving the person seeking reimbursement time and replacing the complicated expense report filling process. This eliminates the need for the person seeking reimbursement to log in to the expense report page and manually fill in the expense report content item by item.
[0021] Secondly, this application provides a reimbursement form filling device based on an intelligent agent, which adopts the following technical solution: A reimbursement form filling device based on intelligent agents, comprising: The acquisition module is used to acquire reimbursement requests entered by users through the intelligent agent assistant interface; The intent recognition module is used to input the reimbursement request into a pre-trained large language model for intent recognition, obtain the intent information of the reimbursement request, and determine the process classification result of the reimbursement request based on the intent information. The process classification result includes querying the business trip task book, generating a reimbursement form based on the business trip task book, and writing invoice information into the reimbursement form details. The execution module is used to trigger the corresponding workflow branch to perform corresponding operations based on the process classification result, obtain the execution result of the reimbursement request, and provide feedback on the execution result of the reimbursement request based on the intelligent agent assistant interface. The operations include calling the database query interface, business plugin interface, or external tool interface.
[0022] Thirdly, this application provides an electronic device that adopts the following technical solution: An electronic device includes a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and executed as described in any of the first aspects, a method for filling out expense reimbursement forms based on an intelligent agent.
[0023] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium storing a computer program capable of being loaded by a processor and executing the agent-based expense reimbursement form filling method described in any of the first aspects.
[0024] Additional aspects and advantages of this application will be set forth in part in the description which follows, and will become apparent from the description or may be learned by practice of this application. Attached Figure Description
[0025] Figure 1 A flowchart illustrating an agent-based expense reimbursement form filling method according to an embodiment of the present invention; Figure 2 This is a schematic diagram of an agent-based expense reimbursement form filling device according to an embodiment of the present invention; Figure 3 This is a schematic diagram of an electronic device provided according to an embodiment of the present invention. Detailed Implementation
[0026] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0027] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.
[0028] This application provides a method for filling out expense reimbursement forms based on intelligent agents. This method can be executed by an electronic device, which can be a server or a mobile terminal device. The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services. The mobile terminal device can be a laptop computer, a desktop computer, etc., but is not limited to these.
[0029] like Figure 1 As shown, a method for filling out expense reimbursement forms based on intelligent agents mainly includes: S1, Obtain the reimbursement request entered by the user through the intelligent agent assistant interface; In this embodiment, the intelligent agent assistant interface can adopt a graphical interface design to facilitate users to quickly input information. Users can initiate reimbursement requests through voice or text input. For example, after returning from a business trip, a user can say to the intelligent agent assistant, "I want to be reimbursed for this business trip expense," or directly enter relevant content in the interface text box. In addition to the intelligent agent assistant interface, the request can also be obtained using the interface of a mobile application.
[0030] S2, input the reimbursement request into a pre-trained large language model for intent recognition, obtain the intent information of the reimbursement request, and determine the process classification result of the reimbursement request based on the intent information. The process classification result includes querying the business trip task book, generating a reimbursement form based on the business trip task book, and writing the invoice information into the reimbursement form details. In this embodiment of the application, the process classification results also include My To-Do List, Approval of Expense Reimbursement Forms, Query of Expense Reimbursement Forms, Expense Reimbursement System Issues, and Policy Inquiries.
[0031] The pre-trained large language model is the Jiusi 72B model, which is obtained by using the LoRA efficient fine-tuning method based on the Qianwen 72B model and has a high semantic understanding capability. When a user enters a reimbursement request, the large model will analyze the keywords and semantic information in the request to determine the intent.
[0032] For example, if a user enters "I want to query my business trip assignment," the large model can recognize this as an intent to query a business trip assignment. Different intents will correspond to different process classification results, such as querying a business trip assignment, generating an expense report based on the business trip assignment, and writing invoice information into the expense report details, etc.
[0033] S3. Based on the process classification result, trigger the workflow branch corresponding to the process classification result to perform the corresponding operation, obtain the execution result of the reimbursement request, and provide feedback on the execution result of the reimbursement request based on the intelligent agent assistant interface. The operation includes calling the database query interface, business plugin interface, or external tool interface.
[0034] In this embodiment, when the process classification result is a query for a business trip assignment, a database query interface is invoked to search for relevant business trip assignment information in the database. The database can be a relational database, such as MySQL, which offers good data management and query performance. Once the result is retrieved, it is fed back to the user through an intelligent assistant interface, such as displaying detailed information about the business trip assignment on the screen.
[0035] Optionally, when the process classification result is "generate expense report based on business trip assignment", the step of triggering the corresponding workflow branch to perform the corresponding operation based on the process classification result includes: Obtain the user's login information and business trip assignment number; Determine whether the user's login information matches the traveler's identity corresponding to the business trip assignment number; If the user's login information matches the traveler's identity corresponding to the business trip assignment number, then the business trip information is extracted from the travel application data corresponding to the business trip assignment. The business trip information includes the project, business trip time, and destination. Call the expense report draft generation interface, pass in the business trip information as a parameter, generate an expense report draft and store it in the database, and provide the expense report number corresponding to the expense report draft based on the intelligent agent assistant interface.
[0036] In this embodiment of the application, for example, after entering the expense report generation branch, the workflow first executes the authentication logic through the business plugin node. This node obtains the current system's logged-in user information (such as user ID: EMP10001) and the business trip assignment number (BJ24071309832) extracted from the user's request. Subsequently, the plugin node calls the business trip assignment query interface, queries the travel application form (travel_applications) based on the number BJ24071309832, and obtains the traveler ID field of the assignment record.
[0037] The system compares the currently logged-in user ID (EMP10001) with the queried traveler ID. If they do not match, the assistant will immediately report "You are not authorized to operate this travel assignment." If the verification is successful, the system will continue to extract key travel information from the travel assignment data, including: the associated project code (PROJ-2024-RD01), travel start time (2025-08-01), travel end time (2025-08-05), departure city (Beijing), and destination city (Shanghai).
[0038] After successfully extracting the business trip information, the workflow engine calls the expense report draft generation interface. This interface receives the extracted information as input parameters and executes the creation logic in the background. The interface first inserts a new record into the expense report draft table (reimbursement_drafts), filling in the project code, start and end dates, city information, etc., and initializing the status to "draft". Upon successful generation, the interface returns the automatically generated expense report draft number (e.g., DRAFT-20250806516). Finally, the workflow engine encapsulates this result into a natural language description and provides feedback to the user through the intelligent agent assistant interface: "Hello, an expense report draft has been successfully generated based on the business trip task book BJ24071309832, with the number: DRAFT-20250806516. You can supplement the invoice or submit for approval at any time." Furthermore, when retrieving business trip information, if it's discovered that the trip involves two projects simultaneously, the generation logic can be enhanced with intelligent cost-sharing. This intelligent cost-sharing logic can automatically allocate the total reimbursement amount (such as accommodation fees and travel allowances) among multiple projects based on preset rules (e.g., the proportion of business trip days) or historical cost-sharing habits. The generated draft reimbursement form will clearly list the project to which each expense belongs and the allocated amount, prompting the user for confirmation.
[0039] During the draft generation stage, a cost standard library can be further integrated for pre-filling and pre-checking. When calling the expense report draft generation interface, in addition to basic business trip information, the system will also retrieve the corresponding accommodation cost standard (e.g., 500 yuan per night in Beijing) and transportation cost from the standard library based on the user's job level (e.g., "technical staff") and destination city. Then, it will automatically create an "accommodation cost" detail for the draft, using the standard amount as the pre-filled value. At the same time, a pre-check will be performed. If the cost of a booked hotel exceeds this standard, an overspending warning will be issued to the user at the same time as the draft is generated, prompting them to prepare explanatory materials in advance.
[0040] Optionally, when the process classification result is that invoice information is written to the expense report details, the step of triggering the corresponding workflow branch to perform the corresponding operation based on the process classification result includes: Receive invoice attachment information; Optical character recognition is performed on the invoice attachment information to obtain structured invoice information; The structured invoice information is subjected to compliance verification, which includes verification of duplicate invoice reimbursement, verification of consistency of the invoicing entity, and verification of the authenticity of the invoice. If the verification is successful, the structured invoice information will be filled into the corresponding expense details in the reimbursement form. If the verification fails, the result will be displayed on the AI assistant interface.
[0041] In this embodiment, for example, a user enters the instruction "Put the train ticket photo information I just uploaded into the travel expenses details of the reimbursement form" in the intelligent agent assistant interface, and uploads the corresponding train ticket photo. The intelligent agent assistant interaction front end receives the text instruction and attachment information, encapsulates it into a structured request, and sends it to the backend workflow engine. The large model classification node in the workflow engine performs intent recognition on the instruction, determines its process classification result as "Write invoice information to reimbursement form details", and then triggers the corresponding "Invoice Processing" workflow branch.
[0042] Upon entering the invoice processing branch, the uploaded image is first standardized using the file preprocessing service, including image correction, noise reduction, and sharpness enhancement. The preprocessed image is then fed into the Optical Character Recognition (OCR) engine. This OCR engine is specifically optimized for invoice formats, enabling it to accurately locate and recognize key fields.
[0043] After obtaining the structured invoice information, the system uses the invoice number as the key index to query all historical reimbursement records in the financial database. If a record of a successfully reimbursed invoice with the same number is found, the verification fails, and the system immediately sends a notification to the assistant stating, "This train ticket was already reimbursed on [date], please do not submit it again."
[0044] Next, the passenger name on the train ticket is compared with the name and ID number of the person currently claiming reimbursement. Simultaneously, it is verified whether the issuing entity of the invoice is an recognized entity.
[0045] Finally, the system calls the State Taxation Administration's electronic invoice database interface through the security gateway, submitting key invoice information (invoice code, number, amount, and invoice date) for verification. Only when the official interface returns a "normal" status can the final verification be passed.
[0046] Once all three layers of verification pass, this information is written into the database as a new expense detail. The expense type is automatically matched as "travel and transportation expenses," and information such as amount, date, origin and destination are automatically filled in. Upon success, the intelligent assistant sends a notification to the user: "Success! Train ticket information has been added to the reimbursement form."
[0047] Once an invoice is successfully added to the expense report details, a real-time budget query can be triggered to assess the impact of this expense on the execution rate of the relevant project budget. If the added expense causes the project budget execution rate to exceed a set threshold, a notification will be sent via an assistant: "Friendly reminder: After adding this invoice, the project budget usage has reached the target progress value. Please be aware."
[0048] After filling the structured invoice information into the corresponding expense details of the reimbursement form, the process also includes: Based on preset expense limit rules, it is determined whether the expense items in the expense details of the reimbursement form exceed the budget. The preset expense limit rules include limits on transportation expenses, accommodation expenses, and total project budget for different job levels.
[0049] If any expense item in the expense report exceeds the budget, an overspending reminder will be generated. The overspending reminder will distinguish between personal overspending and project budget overspending. An overspending template will be displayed on the intelligent assistant interface. After the user fills in the overspending template, the completed overspending template will be marked at the approval process node.
[0050] After the structured invoice information is entered into the expense details of the reimbursement form, based on the preset expense limit rules such as the limit for transportation expenses, accommodation expenses, and the total budget limit for the project, it can determine whether the expense items in the expense details of the reimbursement form are over-spending. This can detect overspending in advance, avoid the expense being returned for modification by the approver, and improve the efficiency of the reimbursement process.
[0051] Optional, also includes: In response to a user's withdrawal request, withdraw the expense report submitted by the user before it enters the approval process stage; In response to a user's request for expedited review, an approval reminder notification is sent to the approver based on the intelligent assistant.
[0052] By proactively predicting user expense reports and recommending and displaying draft expense reports when users log in to the intelligent assistant, the draft is converted into an official expense report after user confirmation. This saves the expense report user time and replaces the tedious expense report filling process, eliminating the need for the expense report user to log in to the expense report page and manually fill in the expense report information item by item.
[0053] As an optional implementation of this application, a reimbursement form filling method based on an intelligent agent further includes: In response to booking events related to user travel in the business system, obtain the user's business trip assignment, historical expense data, travel application data, and travel consumption record data; A correlation analysis was conducted on business trip assignment documents, travel application data, and travel expense records to obtain the correlation analysis results; Based on the correlation analysis results, travel application data and travel consumption record data for the same travel event are correlated and integrated to obtain the integration result; Based on the integrated results, historical reimbursement data, and a pre-set recommendation model, predict the user's reimbursement items; Based on the user's reimbursement items and the integrated results, a draft reimbursement form is generated.
[0054] In this embodiment, the correlation analysis determines whether the consumption date falls within the business trip date range. For example, if Xiao Wang's business trip is from September 10th to September 12th, 2025, his flight ticket booked on September 9th, departing on September 10th, and his hotel check-in record on September 11th are both determined to be related to this business trip and thus associated with the same event. Incompatible records (such as consumption during personal vacation) are filtered out.
[0055] After completing the data association, the feature repository was accessed to extract Xiao Wang's historical expense reimbursement data from the past two years, especially the distribution of his reimbursement items. These features (such as "4 out of the last 5 business trip reimbursements are related to item A") were input into a lightweight machine learning recommendation model (e.g., using a gradient boosting tree or logistic regression model). By learning Xiao Wang's reimbursement habits and the project allocation patterns of his department, the model predicts the most likely reimbursement item associated with this business trip and outputs a probability value. Based on the probability values output for each reimbursement item, the user's historical usage frequency of a specific item on the system, the time of the most recent use of the item, the current status of the item, and the items already reimbursed by peers, the overall confidence level of each reimbursement item is determined, and a list of recommended items is arranged in descending order of overall confidence level.
[0056] It should be noted that if the business trip itinerary specifies reimbursement items, those reimbursement items will receive the highest weight and will be directly used as the recommended result.
[0057] In this embodiment of the application, when it is detected that a user has logged into the intelligent agent assistant, the draft expense report is recommended and displayed on the interface of the intelligent agent assistant. In response to the user's confirmation instruction for the draft expense report, the draft expense report is converted into an official expense report.
[0058] In the above implementation, the intelligent agent assistant will proactively send a heartbeat detection or user online event to the backend service. Upon receiving the event, the backend service immediately queries the expense report draft library to check if there is a draft under Xiao Wang's name with a status of "pending confirmation". If it exists, a highlighted message will be pushed through the intelligent agent assistant interface: "We have discovered that you recently completed a business trip to Shanghai, and an expense report draft has been pre-generated for you."
[0059] When a user has multiple draft expense reports awaiting confirmation (such as a travel expense draft and a purchase expense draft), the display priority of each draft can be calculated based on multi-dimensional context. Sorting factors may include: 1) Time urgency: The closer to the scheduled reimbursement deadline, the higher the priority; 2) Amount size: The larger the amount, the more complex the approval process may be, and it should be processed first; 3) Project status: If the associated project is nearing its settlement period, its priority will be increased.
[0060] A weighted algorithm is used to derive a comprehensive priority score, which is then recommended in order in the assistant interface, along with brief reason tags (such as "Deadline Approaching" or "High Amount") to guide users to process applications efficiently.
[0061] This invention can accurately identify and classify the intent of a user's expense reimbursement request, and then trigger corresponding workflow branches to perform operations based on different classifications. It utilizes multiple interfaces to meet different business needs and provides feedback on the execution results through an intelligent assistant interface. This can save employees' expense reimbursement form filling time, improve the efficiency of the expense reimbursement process, avoid errors and compliance risks from manual operations, and enhance the user experience.
[0062] Figure 2 A schematic diagram of a reimbursement form filling device 200 based on an intelligent agent is shown.
[0063] like Figure 2 As shown, a reimbursement form filling device 200 based on an intelligent agent mainly includes: The acquisition module 201 is used to acquire the reimbursement request entered by the user through the intelligent agent assistant interface; The intent recognition module 202 is used to input the reimbursement request into a pre-trained large language model for intent recognition, obtain the intent information of the reimbursement request, and determine the process classification result of the reimbursement request based on the intent information. The process classification result includes querying the business trip task book, generating a reimbursement form based on the business trip task book, and writing invoice information into the reimbursement form details. The execution module 203 is used to trigger the corresponding operation of the workflow branch corresponding to the process classification result based on the process classification result, obtain the execution result of the reimbursement request, and provide feedback on the execution result of the reimbursement request based on the intelligent agent assistant interface. The operation includes calling the database query interface, business plugin interface or external tool interface.
[0064] In one example, the module in any of the above devices may be one or more integrated circuits configured to implement the above methods, such as one or more application-specific integrated circuits (ASICs), or one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.
[0065] For example, when modules in a device can be implemented via a processing element scheduler, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor capable of calling programs. Alternatively, these modules can be integrated together as a system-on-a-chip (SOC).
[0066] In this application, various objects such as messages / information / devices / network elements / systems / apparatus / actions / operations / processes / concepts may be named. It is understood that these specific names do not constitute a limitation on the relevant objects. The names may be changed depending on the scenario, context, or usage habits. The understanding of the technical meaning of the technical terms in this application should be mainly determined from their functions and technical effects embodied / performed in the technical solution.
[0067] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0068] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.
[0069] Figure 3 This is a structural block diagram of an electronic device 300 according to an embodiment of this application.
[0070] like Figure 3As shown, the electronic device 300 includes a processor 301 and a memory 302, and may further include one or more of an information input / output (I / O) interface 303, a communication component 304, and a communication bus 305.
[0071] The processor 301 controls the overall operation of the electronic device 300 to complete all or part of the steps in the agent-based expense reimbursement form filling method described above. The memory 302 stores various types of data to support the operation of the electronic device 300. This data may include, for example, instructions for any application or method operating on the electronic device 300, as well as application-related data. The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as one or more of Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.
[0072] I / O interface 303 provides an interface between processor 301 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 304 is used to test wired or wireless communication between electronic device 300 and other devices. Wireless communication includes Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 304 may include a Wi-Fi component, a Bluetooth component, and an NFC component.
[0073] The communication bus 305 may include a path for transmitting information between the aforementioned components. The communication bus 305 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 305 may be divided into an address bus, a data bus, a control bus, etc.
[0074] The electronic device 300 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to execute the agent-based expense reimbursement form filling method given in the above embodiments.
[0075] The following describes the computer-readable storage medium provided in the embodiments of this application. The computer-readable storage medium described below can be referred to in correspondence with the agent-based reimbursement form filling method described above.
[0076] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described agent-based expense reimbursement form filling method.
[0077] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0078] The terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0079] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing application concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions claimed in this application.
Claims
1. A method for filling out expense reimbursement forms based on intelligent agents, characterized in that, include: Obtain the reimbursement request entered by the user through the intelligent agent assistant interface; The reimbursement request is input into a pre-trained large language model for intent recognition to obtain intent information of the reimbursement request. Based on the intent information, the process classification result of the reimbursement request is determined. The process classification result includes querying the business trip task book, generating a reimbursement form based on the business trip task book, and writing invoice information into the reimbursement form details. Based on the process classification result, the workflow branch corresponding to the process classification result is triggered to perform the corresponding operation, and the execution result of the reimbursement request is obtained. The execution result of the reimbursement request is fed back based on the intelligent agent assistant interface. The operation includes calling the database query interface, business plugin interface or external tool interface.
2. The method for filling out expense reimbursement forms based on intelligent agents according to claim 1, characterized in that, When the process classification result is "generate expense report based on business trip assignment", the step of triggering the corresponding workflow branch to perform the corresponding operation based on the process classification result includes: Obtain the user's login information and business trip assignment number; Determine whether the user's login information matches the traveler's identity corresponding to the business trip assignment number; If the user's login information matches the traveler's identity corresponding to the business trip assignment number, then the business trip information is extracted from the travel application data corresponding to the business trip assignment. The business trip information includes the project, business trip time, and destination. Call the expense report draft generation interface, pass in the business trip information as a parameter, generate an expense report draft and store it in the database, and provide the expense report number corresponding to the expense report draft based on the intelligent agent assistant interface.
3. The method for filling out expense reimbursement forms based on intelligent agents according to claim 1, characterized in that, When the process classification result indicates that invoice information is written to the expense report details, the step of triggering the corresponding workflow branch based on the process classification result to perform the corresponding operation includes: Receive invoice attachment information; Optical character recognition is performed on the invoice attachment information to obtain structured invoice information; The structured invoice information is subjected to compliance verification, which includes verification of duplicate invoice reimbursement, verification of consistency of the invoice issuer, and verification of the authenticity of the invoice. If the verification is successful, the structured invoice information will be filled into the corresponding expense details in the reimbursement form. If the verification fails, the result will be displayed on the AI assistant interface.
4. The reimbursement form filling method based on an intelligent agent according to claim 3, characterized in that, After filling the structured invoice information into the corresponding expense details of the reimbursement form, the process also includes: Based on preset expense limit rules, it is determined whether the expense items in the expense details of the reimbursement form exceed the budget. The preset expense limit rules include limits on transportation expenses, accommodation expenses, and total project budget based on job level. If any expense item in the expense report exceeds the budget, an overspending reminder will be generated. The overspending reminder will distinguish between personal overspending and project budget overspending. An overspending template will be displayed on the intelligent assistant interface. After the user fills in the overspending template, the completed overspending template will be marked at the approval process node.
5. The reimbursement form filling method based on an intelligent agent according to claim 4, characterized in that, Also includes: In response to a user's withdrawal request, withdraw the expense report submitted by the user before it enters the approval process stage; In response to a user's request for expedited review, an approval reminder notification is sent to the approver based on the intelligent assistant.
6. The method for filling out expense reimbursement forms based on intelligent agents according to claim 1, characterized in that, Also includes: In response to booking events related to user travel in the business system, obtain the user's business trip assignment, historical expense data, travel application data, and travel consumption record data; A correlation analysis was conducted on business trip assignment documents, travel application data, and travel expense records to obtain the correlation analysis results; Based on the correlation analysis results, travel application data and travel consumption record data for the same travel event are correlated and integrated to obtain the integration result; Based on the integrated results, historical reimbursement data, and a pre-set recommendation model, predict the user's reimbursement items; Based on the user's reimbursement items and the integrated results, a draft reimbursement form is generated.
7. The reimbursement form filling method based on an intelligent agent according to claim 6, characterized in that, Also includes: When a user is detected to have logged into the intelligent agent assistant, the draft expense report is recommended and displayed on the intelligent agent assistant interface. In response to the user's confirmation instruction for the draft expense report, the draft expense report is converted into an official expense report.
8. A reimbursement form filling device based on an intelligent agent, characterized in that, include: The acquisition module is used to acquire reimbursement requests entered by users through the intelligent agent assistant interface; The intent recognition module is used to input the reimbursement request into a pre-trained large language model for intent recognition, obtain the intent information of the reimbursement request, and determine the process classification result of the reimbursement request based on the intent information. The process classification result includes querying the business trip task book, generating a reimbursement form based on the business trip task book, and writing invoice information into the reimbursement form details. The execution module is used to trigger the corresponding workflow branch to perform corresponding operations based on the process classification result, obtain the execution result of the reimbursement request, and provide feedback on the execution result of the reimbursement request based on the intelligent agent assistant interface. The operations include calling the database query interface, business plugin interface, or external tool interface.
9. An electronic device, characterized in that, Includes a processor, which is coupled to a memory; The processor is configured to execute a computer program stored in the memory to cause the electronic device to perform the method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, Includes a computer program or instructions that, when run on a computer, cause the computer to perform the method as described in any one of claims 1-7.