Business travel reimbursement management system

Abstract

The invention discloses a business travel reimbursement management system, and relates to the technical field of business travel management, and the system comprises the following components: a data collection module which collects business travel demands and historical business travel records of a user, analyzes the relation between data through an algorithm model, and displays an analysis result to the user and an enterprise in a visual manner, according to the invention, by integrating the business travel related data and combining with the data cleaning and statistical analysis technology, the highly personalized business travel planning service is provided for the user, the historical travel data of the user is classified by applying the K-means algorithm, different user groups and travel modes thereof are accurately identified, and the user experience is improved. According to the method, flight, hotel and travel destination recommendation and association rule mining algorithms which better meet the actual requirements of users are provided for the users, and association rules between travel behaviors of the users and expenses are deeply mined, so that more accurate cost budget and control basis are provided for enterprises, unnecessary expenses are effectively avoided, and the user experience is improved. And the financial management efficiency of enterprises is improved.

Description

Technical Field

[0001] The present invention relates to the field of travel management, and in particular to a travel reimbursement management system. Background Art

[0002] Traditional travel expense management systems often lack an in-depth understanding and intelligent response to users' personalized needs. They can only provide basic itinerary planning and expense budgeting functions, but are unable to make intelligent recommendations and personalized services based on users' travel history, preferences, and travel patterns. As a result, users need to spend a lot of time screening and comparing information when planning their trips, reducing work efficiency and user experience.

[0003] Secondly, the traditional travel expense management system lacks accuracy and dynamism in expense budgeting and control. It can only budget expenses based on fixed budget templates and experience values, and cannot be dynamically adjusted according to the user's actual needs and itinerary changes. It also lacks timely and effective early warning and reminder mechanisms for over-budget risks and non-compliant expense expenditures, which increases the company's financial risks and management costs.

[0004] Finally, the traditional travel expense management system has problems with cumbersome procedures and low efficiency in reimbursement application and processing. Users need to manually fill out reimbursement applications, upload invoices and vouchers, and complete the reimbursement process only after approval. This not only increases the user's workload, but also prolongs the reimbursement cycle, affecting the company's capital flow and operational efficiency. Due to the subjectivity and error rate of manual review, it also increases the company's management risk.

[0005] Therefore, a travel expense management system is developed to solve the above problems and improve the company's travel experience and reimbursement management efficiency. Summary of the invention

[0006] The purpose of the present invention is to make up for the shortcomings of the prior art and provide a travel reimbursement management system, which can achieve accurate classification of user data, identification of travel modes and personalized recommendations of flights, hotels and travel destinations by applying K-means algorithm, association rule mining and collaborative filtering technology. It not only provides intelligent itinerary planning and expense budgeting and control functions, but also can predict budget adjustment needs, and has budget threshold warning and expense control suggestions. It supports users to submit reimbursement applications, automatically generates reimbursement forms after approval, and displays itinerary planning, expense budgeting and reimbursement application information through the user interface, allowing users to edit and adjust. It also has an effect evaluation module to support user feedback to achieve continuous optimization and improvement.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: a travel reimbursement management system, which includes the following components:

[0008] Data collection module: collect users' business trip needs and historical business trip records, clean the collected raw data, apply K-means algorithm to classify users' historical travel data, identify different user groups and their travel patterns, use association rule mining algorithm to mine the association rules between users' travel behavior and expense expenditure, and use collaborative filtering to recommend flights, hotels and travel destinations of interest to users based on users' historical travel data and preferences. Use data visualization technology to display the analysis results to users and enterprises in an intuitive way for easy understanding and use;

[0009] Intelligent itinerary planning module: uses artificial intelligence algorithms to automatically plan the user's entire business trip itinerary, taking into account multi-dimensional factors during the planning process, and providing users with a variety of itinerary options for selection. Users can also customize itinerary settings according to their own needs;

[0010] Cost budget and control module: Generates detailed cost budget for users according to each link of the trip, and can dynamically adjust the budget to ensure the accuracy and effectiveness of the budget. It also issues early warning and reminders for any risk of over-budget or non-compliant expenditures.

[0011] Reimbursement application and processing module: submit travel expense reimbursement applications, upload relevant invoices and vouchers, automatically review and process reimbursement applications, automatically generate reimbursement forms and vouchers after review, and notify users to confirm and collect;

[0012] User interface and interaction module: responsible for providing a user-friendly interface and interaction method, enabling users to easily view information on itinerary planning, expense budgeting, and reimbursement applications, and perform related operations.

[0013] Furthermore, the K-means algorithm is applied to classify the historical travel data of users, identify different user groups and their travel modes, collect the historical travel data of users, select features related to the travel mode, and for each user data point, calculate the Euclidean distance between it and the K cluster centers. The Euclidean distance calculation formula is: Where: d(x i ,μ k ) is the data point x i With cluster center μ k The distance between them, d is the characteristic dimension of the data, and They are x i and μ k The j-th eigenvalue of , assigns the data point to the cluster center closest to it, and redistributes the mean of each cluster as the new cluster center. The mean calculation formula is: Where: n k is the number of all user data points, xi are clustered user data points. Repeat the above steps until the maximum number of iterations is reached. Analyze the user data points in each cluster and extract the common features of these users to identify different user groups and their travel patterns.

[0014] Furthermore, the association rule mining algorithm is used to mine the association rules between user travel behavior and expense expenditure, collect user travel behavior data and corresponding expense expenditure data, form an original data set D, define an item set, and the item set includes various features of user travel behavior and expense expenditure. For each item set X in the data set, its support in the data set is calculated, and the support calculation formula is: Where: count(X) is the number of times the item set X appears in the data set D, |D| is the total number of transactions in the data set D, set the minimum support threshold, use the Apriori algorithm to generate frequent item sets based on the support of the item set and the minimum support threshold, and for each frequent item set X and Y, calculate the confidence of the association rule X→Y. The confidence calculation formula is: Set a minimum confidence threshold, filter out association rules with confidence greater than or equal to the minimum confidence threshold, and output the filtered strong association rules as strong association rules. The mined association rules will be used in the fields of user travel behavior analysis, cost prediction, and recommendation systems to improve user experience and service quality.

[0015] Furthermore, the collaborative filtering is used to recommend flights, hotels and travel destinations of interest to users, collect the user's historical flight, hotel and travel destination booking data, pre-process the data to ensure the accuracy and consistency of the data, and construct a user-item rating matrix R, where: R(u,i) represents the preference of user u for item i, and cosine similarity is selected to calculate the similarity between users. The cosine similarity formula is: Where: I w is the set of items that users u and v have evaluated together, and The average ratings of users u and v are based on user similarity, and the k most similar users are found for the target user u. Then, the ratings of the target user for unevaluated items are predicted based on the preferences of these neighbor users. A recommendation list is generated according to the ratings, and the generated recommendation list is displayed to the user.

[0016] Furthermore, the intelligent itinerary planning module collects travel-related data, pre-processes the collected data to ensure data quality and consistency, uses machine learning algorithms to analyze users' historical data and preferences, builds user portraits, and uses graph search algorithms to determine routes based on user portraits. It preliminarily plans an itinerary that meets user needs, uses linear programming methods to optimize the allocation of resources in the itinerary to find the most cost-effective solution, uses deep learning models to further refine the recommendation system, evaluates risks that may be encountered in the itinerary, and dynamically adjusts the itinerary based on user feedback and real-time data. It uses visualization technology to present the itinerary plan to the user in an intuitive manner, and the user views the recommended itinerary plan and customizes it according to their own needs.

[0017] Furthermore, the selection uses the linear programming method to optimize the allocation of resources in the trip to find the most cost-effective solution. Let the decision variables x1, x2, ..., x n , represents the number of days of accommodation and mode of transportation in the trip, determines the objective function Z, and the linear expression for minimizing the total cost is: Z = c1x1 + c2x2 + ... + c n x n , where: c i is the decision variable x i The unit cost of resources is: The resource allocation constraints are: budget constraint: a 11 x1+a 12 x2+…+a 1n x n ≤, time constraint: a 21 x1+a 22 x2+…+a 2n x n ≤b2, other resource constraints: a m1 x1+a m2 x2+…+a mn x n ≤ b m , the decision variables need to satisfy the non-negativity constraint, x i ≥0, the objective function and constraints are combined to form a complete linear programming model: MinimizeZ=c T xSubjectto:Ax≤bx≥0, where: c is the objective function coefficient vector, A is the constraint matrix, b is the right-hand constant vector of the constraints, and x is the decision variable vector. Use a linear programming solver to solve the above linear programming model, check whether the solution meets all constraints, ensure the feasibility of the solution, analyze the solution results, determine the optimal resource allocation plan, and evaluate its cost-effectiveness.

[0018] Furthermore, the expense budget and control module obtains information on each link of the itinerary from the intelligent itinerary planning module, collects relevant market price information to estimate the cost of each link, generates a total expense budget, monitors changes in market prices in real time, collects the latest market price information, considers the user's actual expenditure, uses a random forest algorithm to predict the need for budget adjustment, audits expense expenditures to ensure that expense expenditures are in line with the budget and company policies, sets up an early warning mechanism, issues an early warning when expense expenditures approach the budget threshold, and provides expense control suggestions to users based on the expense analysis results.

[0019] Furthermore, the steps of using random forest to predict the need for budget adjustment are:

[0020] (1) Extract features related to budget adjustment from the preprocessed data and transform the features;

[0021] (2) Set the parameters of the forest model, use the labeled historical data to train the random forest model, and finally obtain the prediction results through voting;

[0022] (3) Prediction results of random forest It is determined by majority voting on the prediction results of all decision trees. The mathematical expression is: Among them: mode represents the mode, the category with the most occurrences;

[0023] (4) When there is new market price data, actual expenditure data, and expected itinerary changes, they are input into the trained random forest model, and the random forest outputs a physical result indicating whether the budget needs to be adjusted;

[0024] (5) The administrator looks at the model prediction results and reviews and adjusts them based on actual conditions.

[0025] Furthermore, the user of the reimbursement application and processing module logs in to the system and enters the reimbursement application interface. The user fills in the reimbursement application, uploads relevant invoices and receipts, and submits the reimbursement application after confirming that the information is correct. The system verifies whether the information of the reimbursement application is complete. If the reimbursement application passes the system review, it will automatically generate a reimbursement form and voucher, and notify the user to confirm and collect it. If the reimbursement application fails the system review, a notification of failure will be generated, and the reason for failure will be explained. The user can modify the application according to the notification and resubmit the application. After receiving the reimbursement form and voucher automatically generated by the system, the user will check and confirm it. After the user confirms that it is correct, he / she can collect the reimbursement according to the company's prescribed procedures.

[0026] Furthermore, the user of the effect evaluation module logs in through the login interface, and according to the user's identity and authority, loads and displays the itinerary planning list, which contains information on the starting time, destination, and mode of transportation of each itinerary, allowing the user to edit existing itineraries, such as changing the departure time, destination, and hotel, and displays detailed information on the expense budget, including the budget amount and actual expenditure of each expense, allowing the user to adjust the budget, increase or decrease the budget for a certain expense, and displays a list of reimbursement applications, including information on the submission time and status of each application, allowing the user to click on a reimbursement application to view detailed review progress and results, allowing the user to make flexible adjustments based on actual conditions, providing detailed help documents to answer questions encountered by users during use, and allowing the user to make comments and suggestions on the system for continuous improvement.

[0027] Compared with the existing technology, the travel expense management system has the following beneficial effects:

[0028] 1. The present invention integrates travel-related data and combines data cleaning and statistical analysis technology to provide users with highly personalized travel planning services. It applies the K-means algorithm to classify users' historical travel data, accurately identify different user groups and their travel patterns, and provide users with flight, hotel and travel destination recommendations that are more in line with their actual needs. The association rule mining algorithm deeply mines the association rules between user travel behavior and expense expenditure, thereby providing enterprises with more accurate expense budgeting and control basis, effectively avoiding unnecessary expense expenditure and improving the financial management efficiency of enterprises.

[0029] 2. The intelligent itinerary planning module of the present invention adopts an artificial intelligence algorithm to automatically plan the user's entire business trip itinerary, providing users with a variety of itinerary options to choose from, greatly improving the user's travel experience. The expense budget and control module can generate a refined expense budget for the user based on each link of the itinerary, and has the function of dynamically adjusting the budget, ensuring the accuracy and effectiveness of the budget. In terms of reimbursement application and processing, it can automatically review and process reimbursement applications, reducing the tediousness and errors of manual review and improving work efficiency. The user-friendly interface and interactive method enable users to easily view itinerary planning, expense budget, and reimbursement application information, and perform related operations, further improving the system's ease of use and user satisfaction.

[0030] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0032] Figure 1 This is a flow chart of the travel expense management system. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] Embodiment 1:

[0035] This embodiment provides a travel reimbursement management system, which integrates multiple functions including data collection, intelligent itinerary planning, expense budgeting and control, reimbursement application and processing, and effect evaluation. The intelligent itinerary planning module uses multiple data sources, machine learning, and graph search algorithms to provide users with diversified itinerary plans. The expense budgeting and control module combines market prices and random forest algorithms to accurately predict budget adjustment needs to ensure that expense expenditures are in line with the budget and company policies. The reimbursement application processing module supports online filling and submission of reimbursement applications, realizes automatic verification of information and automatic generation of reimbursement forms, greatly improves the efficiency and accuracy of reimbursement, and also has an effect evaluation function, which continuously optimizes system functions and interface layout by collecting user feedback.

[0036] The data collection module collects users’ business trip needs and historical business trip records, including information on flights, hotels, travel destinations, and expenses. It cleans the collected raw data, removes duplicate, erroneous, and irrelevant data, and selects features related to travel modes, such as travel time, travel distance, and destination type. For each user data point, it calculates its Euclidean distance to the K cluster centers. The Euclidean distance calculation formula is: And assign the data points to the nearest cluster center, iteratively update the mean of each cluster as the new cluster center, the mean calculation formula is: Until the maximum number of iterations is reached or the cluster center no longer changes significantly, analyze the user data points in each cluster, extract common features, identify different user groups and their travel modes, collect user travel behavior data and corresponding expense data, form the original data set D, define the item set, including the various features and expense of user travel behavior, calculate the support of each item set in the data set, and the support calculation formula is: And set the minimum support threshold, use the Apriori algorithm to generate frequent item sets, calculate the confidence of the association rules between each frequent item set, and the confidence calculation formula is: And set the minimum confidence threshold, filter out the association rules with confidence greater than or equal to the minimum confidence threshold, and output them as strong association rules. Collect the user's historical flight, hotel or travel destination booking data, and preprocess them to build the user-item rating matrix R, where R represents the preference of user u for item i. Use cosine similarity to calculate the similarity between users. The cosine similarity formula is: And find the k most similar users for the target user, and predict the target user's rating for the unrated items based on the preferences of neighbor users. The calculation formula for the predicted rating pred(u,i) is: A recommendation list is generated and displayed to users. K-means classification results, association rule mining results and collaborative filtering recommendation results are intuitively displayed to users and businesses in the form of charts and images. An interactive interface is provided to allow users and businesses to query, filter and analyze the displayed results.

[0037] The intelligent itinerary planning module collects data closely related to travel from multiple data sources, ensures data quality and consistency through preprocessing, uses machine learning algorithms to analyze users' historical data and preferences, and builds detailed user portraits. Based on user needs and preferences, the module uses a graph search algorithm to plan a travel route that meets user expectations and preliminarily formulates an itinerary plan. In order to ensure the cost-effectiveness of the itinerary, the linear programming method is further used to set decision variables to represent the number of days of accommodation and transportation methods, and to construct an objective function that minimizes the total cost. During the optimization process, budget constraints, time constraints and other resource constraints are considered to ensure that the decision variables meet non-negative conditions. The model is solved by a linear programming solver to obtain the optimal resource allocation plan and evaluate its cost-effectiveness. Throughout the process, a deep learning model will also be used to evaluate potential risks in the itinerary and dynamically adjust it based on user feedback and real-time data. The itinerary plan will be presented to the user in an intuitive way for them to view and customize.

[0038] The cost budget and control module obtains information on each link of the itinerary from the intelligent itinerary planning module, such as accommodation, transportation, and catering, and collects relevant market price information, such as hotel prices, air ticket prices, and catering prices, to ensure the accuracy and real-time nature of the budget. Based on the itinerary information and market prices obtained, various factors such as seasonal price fluctuations, promotional activities, and user preferences are considered to estimate the cost of each link and generate a total cost budget. The module monitors changes in market prices in real time and collects the latest market price information. It also considers the user's actual expenditure and uses the random forest algorithm to predict the need for budget adjustment. Features related to budget adjustment are extracted from the preprocessed data, such as historical expenditure data, market price change rate, and itinerary changes. The extracted features are converted as necessary to meet the requirements of the random forest model, and the parameters of the random forest model are set. The model is trained using labeled historical data. The mathematical expression is: When there is new market price data, actual expenditure data and expected itinerary changes, they are input into the trained model and the prediction results are obtained by voting. The prediction result is usually a category, indicating whether the budget needs to be adjusted. The administrator views the results of the model prediction and reviews them in light of the actual situation. If the prediction results show that the budget needs to be adjusted, the administrator will adjust the budget according to the actual situation and update the relevant information in the expense budget and control module. The user's expense expenditure is reviewed to ensure that the expenditure is in line with the budget and company policy, and an early warning mechanism is set up. When the expense expenditure is close to the budget threshold and there is a risk of violation, an early warning is issued to the user and administrator in advance. In order to facilitate users and administrators to view the expense budget and control situation, the module provides a visual presentation function.

[0039] In the reimbursement application and processing module, the user logs in by entering the user name and password through the system login interface. On the reimbursement application interface, the user fills in detailed reimbursement information, including the date of the business trip, the location of the business trip, and the type of expense. After filling in the reimbursement application, the user uploads the relevant invoices and receipts. After confirming that the information is correct, the user clicks the Submit button to submit the reimbursement application. The system will review the reimbursement application submitted by the user. If the reimbursement application passes the system review, the system will automatically generate the reimbursement form and voucher, and notify the user to confirm and collect it. If the reimbursement application fails the system review, the system will generate a notification of failure and explain the reason for failure. After the user receives the reimbursement form and voucher automatically generated by the system, he needs to check and confirm it. After the user confirms that it is correct, he can collect the reimbursement according to the company's prescribed procedures.

[0040] Effect evaluation module, the user logs in through the login interface, the system collects the functional requirements of different user roles for the system, such as itinerary planning, expense budgeting, reimbursement application, itinerary planning, according to the user's identity and authority, loads and displays the itinerary planning list, the list contains the starting time, destination, and transportation information of each itinerary, allowing the user to click on a certain itinerary to view and edit, display detailed information on the expense budget, including the budget amount and actual expenditure of each expense, allowing the user to adjust the budget as needed, such as increasing or decreasing the budget for a certain expense, display a list of reimbursement applications, including the submission time and status information of each application, allowing the user to click on a reimbursement application to view the detailed review progress and results, allowing the user to edit existing itineraries, such as changing the departure time, destination, and hotel, and provide the expense budget adjustment function, allowing users to make flexible adjustments according to actual conditions, provide detailed help documents, answer questions encountered by users during use, and allow users to make comments and suggestions on the system, based on user feedback and actual needs, so as to continuously improve and optimize.

[0041] Embodiment 2:

[0042] This embodiment provides a travel expense management system, with particular emphasis on the intelligent itinerary planning module:

[0043] Collect travel-related data from multiple data sources, including flight databases, hotel reservation systems, map services, and weather forecasts. These data include flight information, hotel prices, transportation methods, and destination weather. Perform preprocessing operations on the collected data, including data cleaning, format conversion, and outlier processing, to ensure data quality and consistency. Through data mining and analysis, we can extract key information about users' travel habits, budget constraints, and comfort requirements, and build user portraits, which will serve as an important reference for subsequent itinerary planning.

[0044] Based on user needs and preferences, we use graph search algorithms to consider various factors, such as distance, time, and transportation methods to find the route that best meets user needs. We will preliminarily plan an itinerary that meets user needs, including the number of days of accommodation and transportation methods.

[0045] Based on the preliminary planning, the linear programming method is used to optimize the allocation of resources in the itinerary to find the most cost-effective solution. Let the decision variables x1, x2, …, x n , represents the number of days of accommodation and mode of transportation in the trip, and the linear expression that minimizes the total cost: Z = c1x1+c2x2+…+c n x n , determine the constraints for resource allocation, including budget constraints, time constraints and other resource constraints. The constraints can be expressed as:

[0046] Budget constraint: a 11 x1+a 12 x2+…+a 1n x n ≤, time constraint: a 21 x1+a 22 x2+…+a 2n x n ≤b2, other resource constraints: a m1 x1+a m2 x2+…+a mn x n ≤ b m , the decision variables need to satisfy the non-negativity constraint, x i ≥0, combine the objective function and constraints to form a complete linear programming model: MinimizeZ=c T xSubjectto:Ax≤bx≥0, where: c is the objective function coefficient vector, A is the constraint matrix, b is the right-hand constant vector of the constraints, and x is the decision variable vector. Use a linear programming solver to solve the above linear programming model, check whether the solution meets all constraints, ensure the feasibility of the solution, analyze the solution results, determine the optimal resource allocation plan, and evaluate its cost-effectiveness.

[0047] Use deep learning models to assess the risks that may be encountered during the trip, including weather changes, flight delays, and failed hotel reservations. Deep learning models can predict the probability of these risks by learning historical data and real-time data, and give corresponding risk levels. Based on the risk assessment results, the itinerary can be adjusted accordingly to reduce potential risks.

[0048] The itinerary is dynamically adjusted based on user feedback and real-time data. For example, if the user finds that a certain accommodation location does not meet expectations during the itinerary, the accommodation location or transportation method will be adjusted in real time. The itinerary plan is presented to the user in an intuitive manner using visualization technology, making it easy for the user to view and choose. The user can customize the settings according to their needs. The system will adjust the itinerary plan based on the user's customized settings, and re-optimize resources and re-evaluate risks.

[0049] Embodiment three:

[0050] Based on the first and second embodiments, the specific process steps of the system are as follows:

[0051] 1. The user logs in to the travel expense management system using his / her username and password;

[0052] 2. Create and view itinerary plans, including information on destinations and transportation methods;

[0053] 3. Check and adjust the cost budget for each trip;

[0054] 4. Fill out the reimbursement application, including expense type, amount, and upload relevant vouchers, and submit the application to the review process;

[0055] 5. The system reviews the reimbursement application. Once the review is passed, a reimbursement form is generated and the user is notified;

[0056] 6. The user checks the reimbursement form information and receives the reimbursement;

[0057] 7. Users provide feedback on system usage, and administrators perform daily maintenance and updates of the system.

[0058] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. Travel expense management system, characterized by: The system includes: Data collection module: collect users' business trip needs and historical business trip records, clean the collected raw data, apply K-means algorithm to classify users' historical travel data, identify different user groups and their travel patterns, use association rule mining algorithm to mine the association rules between users' travel behavior and expense expenditure, and use collaborative filtering to recommend flights, hotels and travel destinations of interest to users based on users' historical travel data and preferences. Use data visualization technology to display the analysis results to users and enterprises in an intuitive way for easy understanding and use; Intelligent itinerary planning module: uses artificial intelligence algorithms to automatically plan the user's entire business trip itinerary, taking into account multi-dimensional factors during the planning process, and providing users with a variety of itinerary options for selection. Users can also customize itinerary settings according to their own needs; Cost budget and control module: Generates detailed cost budget for users according to each link of the trip, and can dynamically adjust the budget to ensure the accuracy and effectiveness of the budget. It also issues early warning and reminders for any risk of over-budget or non-compliant expenditures. Reimbursement application and processing module: submit travel expense reimbursement applications, upload relevant invoices and vouchers, automatically review and process reimbursement applications, automatically generate reimbursement forms and vouchers after review, and notify users to confirm and collect; User interface and interaction module: responsible for providing a user-friendly interface and interaction method, enabling users to easily view information on itinerary planning, expense budgeting, and reimbursement applications, and perform related operations.

2. The travel expense management system according to claim 1, characterized in that: The K-means algorithm is applied to classify the historical travel data of users, identify different user groups and their travel modes, collect the historical travel data of users, select the features related to the travel mode, and for each user data point, calculate the Euclidean distance between it and the K cluster centers. The Euclidean distance calculation formula is: in: It is a data point With cluster center The distance between is the characteristic dimension of the data, , assign the data point to the cluster center closest to it, and redistribute the mean of each cluster as the new cluster center. The mean calculation formula is: ,in: is the number of all user data points, are clustered user data points. Repeat the above steps until the maximum number of iterations is reached. Analyze the user data points in each cluster and extract the common features of these users to identify different user groups and their travel patterns.

3. The travel expense management system according to claim 1, characterized in that: The association rule mining algorithm is used to mine the association rules between user travel behavior and expense expenditure, collect user travel behavior data and corresponding expense expenditure data, form an original data set D, define item sets, and the item sets include various features of user travel behavior and expense expenditure. For each item set X in the data set, its support in the data set is calculated. The support calculation formula is: ,in: It is an item set In the dataset The number of times it appears in It is a dataset The total number of transactions in the data set, set the minimum support threshold, use the Apriori algorithm to generate frequent item sets based on the support of the item set and the minimum support threshold, and for each frequent item set X and Y, calculate the confidence of the association rule X→Y. The confidence calculation formula is: , set the minimum confidence threshold, filter out the association rules with confidence greater than or equal to the minimum confidence threshold as strong association rules, output the filtered strong association rules, and use the mined association rules in the fields of user travel behavior analysis, cost prediction, and recommendation system to improve user experience and service quality.

4. The travel expense management system according to claim 1, characterized in that: The collaborative filtering is used to recommend flights, hotels and travel destinations of interest to users, collect the user's historical flight, hotel and travel destination booking data, pre-process the data to ensure the accuracy and consistency of the data, and construct a user-item rating matrix R, where: R Indicates the preference of user u for item i. Cosine similarity is used to calculate the similarity between users. The cosine similarity formula is: ,in: Is a user and A collection of projects that have been reviewed together, and The users are and The average rating of is based on user similarity, the k most similar users are found for the target user u, and then the rating of the target user to the unevaluated items is predicted based on the preferences of these neighbor users, a recommendation list is generated according to the rating, and the generated recommendation list is displayed to the user.

5. The travel expense management system according to claim 1, characterized in that: The intelligent itinerary planning module collects travel-related data, pre-processes the collected data to ensure data quality and consistency, uses machine learning algorithms to analyze users' historical data and preferences, builds user portraits, uses graph search algorithms to determine routes based on user portraits, preliminarily plans itinerary plans that meet user needs, uses linear programming methods to optimize the allocation of resources in the itinerary to find the most cost-effective plan, uses deep learning models to further refine the recommendation system, evaluates risks that may be encountered in the itinerary, dynamically adjusts the itinerary based on user feedback and real-time data, and uses visualization technology to present itinerary plans to users in an intuitive manner. Users view recommended itinerary plans and customize them according to their needs.

6. The travel expense management system according to claim 5, characterized in that: The linear programming method is used to optimize the allocation of resources in the itinerary to find the most cost-effective solution. The decision variables are set , represents the number of days of accommodation and mode of transportation in the trip, and determines the objective function , the linear expression that minimizes the total cost is: ,in: is the decision variable The unit cost of resource allocation is: budget constraint: , Time constraints: , Other resource constraints: , the decision variables need to satisfy the non-negativity constraints, ≥0, the objective function and constraints are combined to form a complete linear programming model: , where: c is the objective function coefficient vector, A is the constraint matrix, b is the right-hand constant vector of the constraints, and x is the decision variable vector. Use the linear programming solver to solve the above linear programming model, check whether the solution meets all the constraints, ensure the feasibility of the solution, analyze the solution results, determine the optimal resource allocation plan, and evaluate its cost-effectiveness.

7. The travel expense management system according to claim 1, characterized in that: The expense budget and control module obtains information on each link of the itinerary from the intelligent itinerary planning module, collects relevant market price information to estimate the cost of each link, generates a total expense budget, monitors changes in market prices in real time, collects the latest market price information, considers the user's actual expenditure, uses a random forest algorithm to predict the need for budget adjustment, audits expense expenditures to ensure that expense expenditures are in line with the budget and company policies, sets up an early warning mechanism, issues an early warning when expense expenditures approach the budget threshold, and provides expense control suggestions to users based on the expense analysis results.

8. The travel expense management system according to claim 7, characterized in that: The steps for using random forest to predict the need for budget adjustment are: (1) Extract features related to budget adjustment from the preprocessed data and transform the features; (2) Set the parameters of the forest model, use the labeled historical data to train the random forest model, and finally obtain the prediction results by voting; (3) Prediction results of random forest It is determined by majority voting on the prediction results of all decision trees. The mathematical expression is: , where: mode represents the mode, the category with the most occurrences; (4) When there is new market price data, actual expenditure data, and expected itinerary changes, they are input into the trained random forest model, and the random forest outputs a physical result indicating whether the budget needs to be adjusted; (5) The administrator looks at the model prediction results and reviews and adjusts them based on actual conditions.

9. The travel expense management system according to claim 1, characterized in that: The user of the reimbursement application and processing module logs in to the system and enters the reimbursement application interface. The user fills in the reimbursement application, uploads relevant invoices and receipts, and submits the reimbursement application after confirming that the information is correct. The system verifies whether the information of the reimbursement application is complete. If the reimbursement application passes the system review, it will automatically generate a reimbursement form and voucher, and notify the user to confirm and collect it. If the reimbursement application fails the system review, a notification of failure will be generated, and the reason for failure will be explained. The user can modify it according to the notification and resubmit the application. After receiving the reimbursement form and voucher automatically generated by the system, the user will check and confirm it. After the user confirms that it is correct, he / she can collect the reimbursement according to the company's prescribed procedures.

10. The travel expense management system according to claim 1, characterized in that: The user of the effect evaluation module logs in through the login interface, and loads and displays the itinerary planning list according to the user's identity and authority. The itinerary planning list contains information about the starting time, destination, and mode of transportation of each itinerary, allowing the user to edit existing itineraries, such as changing the departure time, destination, and hotel, and displays detailed information on the expense budget, including the budget amount and actual expenditure of each expense, allowing the user to adjust the budget, increase or decrease the budget for a certain expense, and displays a list of reimbursement applications, including information on the submission time and status of each application, allowing the user to click on a reimbursement application to view detailed review progress and results, allowing the user to make flexible adjustments based on actual conditions, providing detailed help documents to answer questions encountered by users during use, and allowing the user to make comments and suggestions on the system for continuous improvement.

Images