Intelligent accounting management system and method for business travel reimbursement based on cloud platform

Abstract

The invention discloses an intelligent accounting management system and method for business travel reimbursement based on a cloud platform, and relates to the technical field of business travel reimbursement supervision, and the system comprises a feature processing module, an anomaly analysis module, and a reimbursement processing module. The feature processing module is used for collecting feature parameters of each consumable record and converting attribute type parameters into numeric type parameters through encoding processing to train a travel reimbursement abnormal probability model; the anomaly analysis module is used for monitoring and collecting consumption records when the user is on a business trip in real time, screening the consumption records when the user is on the business trip, and analyzing the anomaly probability and the overlap ratio of each reimbursement record; and the reimbursement processing module performs type division on each reimbursement record through threshold judgment, then performs reimbursement processing and feedback, and analyzes an auditing priority evaluation value of the manual auditing record.

Description

Technical Field

[0001] The present invention relates to the technical field of travel reimbursement supervision, and in particular to a cloud platform-based travel reimbursement intelligent accounting management system and method. Background Art

[0002] Intelligent accounting of travel reimbursement refers to the automation and systematization of the entire process of reviewing corporate personnel travel reimbursement through intelligent analysis of travel reimbursement records. However, the following technical defects exist in the commonly used travel reimbursement review process:

[0003] On the one hand, in smaller-scale data scenarios, electronic tools are generally used to assist manual accounting to review travel reimbursements. The overall process is inefficient and prone to errors, and it is difficult to ensure data security and integration standards. On the other hand, in scenarios where large-scale travel reimbursement data is processed and calculated, accounting and auditing systems based on clustering algorithms or big data analysis are usually used. They have a high degree of data dependence, high data processing volume and complexity, poor flexibility in actual use, and low data sharing. There are loopholes in the rules in the processing of illegal and abnormal reimbursements, and there are problems of people exploiting loopholes to make illegal reimbursements.

[0004] Therefore, a cloud-based travel reimbursement intelligent accounting management system and method is needed to solve the above technical problems. Summary of the invention

[0005] The purpose of the present invention is to provide a cloud platform-based intelligent travel reimbursement accounting management system and method to solve the problems raised in the prior art.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A cloud platform-based intelligent accounting management method for travel reimbursement includes the following analysis steps:

[0008] Step S100: Acquire travel reimbursement history data and reimbursement review results, screen and identify feature parameters of all historical data, perform type conversion on the feature parameters and construct a data set;

[0009] Step S200: construct a logistic regression model and use the data set to train the model to obtain a travel reimbursement abnormality probability model under different feature parameters;

[0010] Step S300: obtaining the consumption records of the user during the travel process through the cloud platform, determining whether each record can be reimbursed, collecting characteristic parameters of the reimbursable records, and using the travel reimbursement abnormal probability model to analyze the abnormal probability of each reimbursable record;

[0011] Step S400: Calculate and analyze the overlap of various reimbursable records based on the characteristic parameters of the reimbursable records;

[0012] Step S500: Perform threshold judgment on the abnormal probability and overlap data of the reimbursable records, classify the processing types of the reimbursable records, and process different types of reimbursable records separately.

[0013] According to the above technical solution, step S100 includes the following contents:

[0014] Collect historical data on travel reimbursement, filter and identify all characteristic parameters that affect the judgment of travel reimbursement anomalies; the characteristic parameters are divided into attribute characteristic parameters and numerical characteristic parameters;

[0015] For attribute-type feature parameters, use one-hot encoding to convert each attribute type in the attribute-type feature parameter into a numerical data format; for any attribute-type feature parameter, analyze all subtypes that may appear in the current feature parameter, set a binary code string with a length equal to the number of all subtypes, and each bit in the code string corresponds to a different subtype; for any subtype in the current feature parameter, set the value of the corresponding bit in the code string to 1, and set the remaining bits to 0, thereby converting all subtypes in the current feature parameter into different numerical data; converting the attribute parameter into the data format of the data-type feature column through one-hot encoding facilitates the direct application of classification parameters during model training and simplifies the data processing link of model training;

[0016] All feature parameters are combined with numerical data and the corresponding travel reimbursement anomaly judgment results to construct a travel reimbursement dataset.

[0017] According to the above technical solution, step S200 includes the following contents:

[0018] A travel reimbursement abnormal probability model is constructed and the travel reimbursement dataset is used for model training. The model is as follows:

[0019]

[0020] Among them, P(Rei Unusual |X) is the abnormal probability of travel reimbursement predicted using the feature parameter combination X, exp() is the exponential function of the natural logarithm, i is the feature parameter number in the feature parameter combination X, n X is the number of characteristic parameters in the characteristic parameter combination X, α0 is the constant parameter of the travel reimbursement abnormal probability model, α i is the weight coefficient of each feature parameter in the feature parameter combination X in the travel reimbursement abnormal probability model, X i is the value of the characteristic parameter i of the travel reimbursement data in the characteristic parameter combination X.

[0021] By constructing a logistic regression model to analyze travel reimbursement anomalies, discrete clustering results are converted into continuous anomaly probabilities, providing a more flexible and rich data foundation and providing data support for subsequent analysis processes.

[0022] According to the above technical solution, step S300 includes the following contents:

[0023] According to the business trip reimbursement standards formulated by the enterprise, the range of various characteristic parameters of the consumption records of users during business trips is limited. When any consumption record of the user during the business trip meets the limited range of all characteristic parameters, the current consumption record is determined to meet the reimbursement standards; by judging the compliance of the user's consumption records, the user's consumption records are initially limited, simplifying the data volume of the subsequent abnormal analysis process;

[0024] Collect the consumption location and user's actual geographic location of each consumption record at the consumption time, and set a geographic orientation deviation threshold; for any consumption record, if the deviation between the consumption location of the consumption record at the consumption time and the user's actual geographic location is greater than the geographic orientation deviation threshold, the current consumption record is judged to lack authenticity and is an abnormal reimbursement record, and authenticity abnormality feedback is sent to the user and the reimbursement reviewer; if the deviation between the consumption location of the consumption record at the consumption time and the user's actual geographic location is less than or equal to the geographic orientation deviation threshold, the current consumption record is judged to have passed the authenticity verification; all consumption records that meet the enterprise reimbursement standards are verified synchronously in time and space again to ensure the authenticity of the user's consumption records and avoid consumption record fraud;

[0025] Determine that consumption records that meet the reimbursement standards and pass authenticity verification are reimbursable records, monitor the user's consumption records when on a business trip in real time through the cloud platform and screen out reimbursable records, use the characteristic parameter identification method to collect the current reimbursable record characteristic parameters, and then predict the abnormal probability of the current reimbursable record based on the travel reimbursement abnormal probability model.

[0026] According to the above technical solution, step S400 includes the following contents:

[0027] Monitor and collect the time intervals of all reimbursable records that have passed authenticity verification, and then calculate the overlap between reimbursable records a and b according to the formula:

[0028]

[0029] Among them, C a，b is the overlap between reimbursable records a and b, Δt is the consumption time interval between reimbursable records a and b, exp() is the exponential function of natural logarithm, m is the attribute characteristic parameter number of the reimbursable record, n is the numerical characteristic parameter number of the reimbursable record, and n mis the number of attribute characteristic parameters of reimbursable records, n n is the number of numerical characteristic parameters of reimbursable records, x a，m is the parameter value of the attribute characteristic parameter m of the reimbursable record a, x b，m is the parameter value of the attribute characteristic parameter m of the reimbursable record b, x a，n is the parameter value of the numerical characteristic parameter n of the reimbursable record a, x b，n is the parameter value of the numerical characteristic parameter n of the reimbursable record b, k m and k n are the evaluation weights of the attribute characteristic parameter m and the numerical characteristic parameter n of the reimbursable record, respectively; |x| is the absolute value of x; flag() is the flag function, when the condition in the brackets is met, the function output is 1, when the condition in the brackets is not met, the function output is 0; max() is the maximum value function, the function output value is the maximum value of all the data in the brackets;

[0030] Combined with the consumption time intervals between various reimbursable records, the overlap of various reimbursable records is analyzed, and the probability of duplicate reimbursement in reimbursement anomalies is further analyzed to ensure the scientificity and accuracy of the final audit results of the overall reimbursement review process.

[0031] According to the above technical solution, step S500 includes the following contents:

[0032] The reimbursable records are divided into automatic reimbursement records, manual review records, and rejected reimbursement records;

[0033] Calculate the overlap between the user's reimbursable records and set the overlap threshold. For any reimbursable record, if there is another reimbursable record with an overlap greater than or equal to the overlap threshold, it is determined that there is a risk of duplicate reimbursement between the two reimbursable records, and both reimbursable records are classified as manual review records; if the overlap between the current reimbursable record and all other reimbursable records is less than the overlap threshold, it is determined that there is no risk of duplicate reimbursement for the current reimbursable record, and subsequent review and analysis of the current reimbursable record is performed;

[0034] The abnormal probability model of travel reimbursement is used to analyze the abnormal probability P of the current reimbursable record, and the manual review boundary [P1, P2] is set; when P<P1, the current reimbursable record is classified as an automatic reimbursement record; when P1≤P≤P2, the current reimbursable record is classified as a manual review record; when P>P2, the current reimbursable record is classified as a rejected reimbursement record; by dividing the user's travel consumption records in detail and dividing each consumption record into different types, the interpretability and operability of the overall process results are improved, and data support is provided for the final accounting results;

[0035] For automatic reimbursement records, the cloud platform will collect statistics on all automatic reimbursement records of users during business trips and automatically reimburse them through the system at the end of the business trip; for rejected reimbursement records, the user will be sent a rejected reimbursement feedback and be reminded to submit paper review materials for reimbursement review at the end of the business trip;

[0036] For manual review records, calculate the review priority evaluation values ​​of all manual review records, and submit manual review applications in descending order according to the calculation results; for any manual review record a, according to the formula:

[0037]

[0038] Among them, L a is the audit priority evaluation value of manual audit record a, T is the audit waiting time of manual audit record a, k1 is the abnormal probability evaluation coefficient of manual audit record a, P a is the abnormal probability of manual review record a, k2 is the duplicate reimbursement evaluation coefficient of manual review record a, j is the reimbursable record number whose overlap with manual review record a is greater than or equal to the overlap threshold, n j is the number of reimbursable records whose overlap with the manually reviewed record a is greater than or equal to the overlap threshold, C a，j It is the overlap between the manually reviewed record a and the reimbursable record j.

[0039] A cloud-based intelligent travel reimbursement accounting management system using a cloud-based intelligent travel reimbursement accounting management method in the above technical solution includes a feature processing module, an abnormality analysis module, and a reimbursement processing module;

[0040] The feature processing module is used to collect feature parameters of various consumable records, and convert the attribute parameters into numerical parameters through encoding processing to train the travel reimbursement abnormality probability model; the abnormality analysis module is used to monitor and collect the consumption records of users when they are on business trips in real time, and screen the users' business trip consumption records, and analyze the abnormal probability and overlap of various reimbursable records; the reimbursement processing module classifies the various reimbursable records into types through threshold judgment, and then performs reimbursement processing and feedback respectively, and analyzes the audit priority evaluation value of the manually reviewed records.

[0041] According to the above technical solution, the feature processing module includes a feature parameter acquisition unit, an attribute encoding unit, and a model data set construction unit;

[0042] The feature parameter collection unit is used to screen and identify all feature parameters that affect the judgment of travel reimbursement anomalies; the attribute encoding unit is used to convert the attribute parameters in the screened feature parameters into numerical parameters using one-hot encoding to participate in subsequent analysis; the model data set construction unit is used to combine all feature parameters with numerical data and corresponding travel reimbursement anomaly judgment results to construct a travel reimbursement data set.

[0043] According to the above technical solution, the abnormal analysis module includes a user consumption record collection unit, a compliance verification unit, an authenticity verification unit, an abnormal reimbursement abnormal probability analysis unit, and a coincidence calculation unit;

[0044] The user consumption record collection unit is used to monitor and collect various consumption records of users during business trips in real time; the compliance verification unit is used to verify whether the various consumption records of users comply with the company's reimbursement standards; the authenticity verification unit is used to analyze the consumption time and place of various consumption records of users to verify the authenticity of various consumption records; the abnormal reimbursement abnormal probability analysis unit uses the reimbursement abnormal probability model to analyze the abnormal probability of various reimbursable records; the overlap calculation unit is used to calculate the overlap data between various reimbursable records.

[0045] According to the above technical solution, the reimbursement processing module includes a reimbursement division unit, an automatic reimbursement unit, a manual review unit, and a reimbursement rejection unit;

[0046] The reimbursement classification unit classifies various reimbursable records into types through threshold judgment; the automatic reimbursement unit counts all automatic reimbursement records of the user during the business trip through the cloud platform and automatically reimburses through the system at the end of the business trip; the manual review unit is used to analyze the review priority of the manual review records and submit manual reimbursement reviews in order from high to low priority; the reimbursement refusal unit sends reimbursement refusal feedback to the user and reminds the user to submit paper review materials for reimbursement review at the end of the business trip.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] In the present invention, the consumption records of business trip users are collected and monitored in real time through a cloud platform, and the travel consumption data is automatically analyzed and calculated to ensure the authenticity and real-time nature of the consumption data, provide a data basis for the automated process of travel reimbursement accounting, and avoid the inefficiency and accounting errors caused by data accumulation at the end of a business trip caused by manual accounting; on the other hand, by constructing a logistic regression model, the classification problem of travel reimbursement anomalies is converted into an abnormal probability analysis problem, and the discrete abnormal classification results are converted into continuous abnormal probability results. The final result is determined by adjusting the abnormal judgment threshold, thereby improving the interpretability and operability of the final travel reimbursement anomaly analysis result; in addition, the travel reimbursement anomaly analysis includes steps such as consumption compliance judgment, time-space authenticity judgment, duplicate reimbursement judgment, and abnormal probability analysis, thereby ensuring the scientificity and accuracy of the final reimbursement accounting result. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of a cloud platform-based intelligent accounting management method for travel reimbursement in the present invention;

[0050] Figure 2 This is an organizational chart of a cloud platform-based travel reimbursement intelligent accounting management system according to the present invention. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Example: See Figure 1-Figure 2 , the present invention provides a technical solution,

[0053] like Figure 1 As shown, the present invention provides a travel expense intelligent accounting management method based on a cloud platform, comprising the following analysis steps:

[0054] Step S100: Acquire travel reimbursement history data and reimbursement review results, screen and identify feature parameters of all historical data, perform type conversion on the feature parameters and construct a data set;

[0055] Step S200: construct a logistic regression model and use the data set to train the model to obtain a travel reimbursement abnormality probability model under different feature parameters;

[0056] Step S300: obtaining the consumption records of the user during the travel process through the cloud platform, determining whether each record can be reimbursed, collecting characteristic parameters of the reimbursable records, and using the travel reimbursement abnormal probability model to analyze the abnormal probability of each reimbursable record;

[0057] Step S400: Calculate and analyze the overlap of various reimbursable records based on the characteristic parameters of the reimbursable records;

[0058] Step S500: Perform threshold judgment on the abnormal probability and overlap data of the reimbursable records, classify the processing types of the reimbursable records, and process different types of reimbursable records separately.

[0059] Step S100 includes the following contents:

[0060] Collect historical data on travel reimbursement, filter and identify all characteristic parameters that affect the judgment of travel reimbursement anomalies; the characteristic parameters are divided into attribute characteristic parameters and numerical characteristic parameters;

[0061] For attribute-type feature parameters, use one-hot encoding to convert each attribute type in the attribute-type feature parameter into a numeric data format; for any attribute-type feature parameter, analyze all possible subtypes in the current feature parameter, set a binary code string with a length equal to the number of all subtypes, and each bit in the code string corresponds to a different subtype; for any subtype in the current feature parameter, set the value of the corresponding bit in the code string to 1, and the remaining bits to 0, thereby converting all subtypes in the current feature parameter into different numeric data;

[0062] In specific implementation, the collected parameters usually include the type of business travelers, consumption location, consumption time, consumption type and consumption amount, etc. For attribute parameters such as the type of business travelers, consumption location and consumption type, all the included types that have appeared in the historical data of corporate travel reimbursement are analyzed. Taking the consumption location as an example, there are three locations in the historical data, namely Beijing, Shanghai and Guangdong. The number of types included in the consumption location parameter is 3. The three types are converted into 100, 010 and 001 respectively using unique hot encoding, and all attribute parameters are further converted into numerical parameters using the above method;

[0063] All feature parameters are combined with numerical data and the corresponding travel reimbursement anomaly judgment results to construct a travel reimbursement dataset.

[0064] Step S200 includes the following contents:

[0065] A travel reimbursement abnormal probability model is constructed and the travel reimbursement dataset is used for model training. The model is as follows:

[0066]

[0067] Among them, P(Rei Unusual |X) is the abnormal probability of travel reimbursement predicted using the feature parameter combination X, exp() is the exponential function of the natural logarithm, i is the feature parameter number in the feature parameter combination X, n X is the number of characteristic parameters in the characteristic parameter combination X, α0 is the constant parameter of the travel reimbursement abnormal probability model, α i is the weight coefficient of each feature parameter in the feature parameter combination X in the travel reimbursement abnormal probability model, X i is the value of the characteristic parameter i of the travel reimbursement data in the characteristic parameter combination X.

[0068] Step S300 includes the following contents:

[0069] According to the business trip reimbursement standards set by the enterprise, the range of various characteristic parameters of the consumption records of the user during the business trip is limited. When any consumption record of the user during the business trip meets the limited range of all characteristic parameters, the current consumption record is determined to meet the reimbursement standards;

[0070] In the specific implementation, the enterprise reimbursement standards are converted into a record screening rule library, and the consumption records of all users during business trips are screened by characteristic parameters. Then, the rules are judged according to the record screening rule library to screen out all consumption records that meet the enterprise reimbursement standards;

[0071] Collect the consumption location and the user's actual geographic location at the time of consumption for each consumption record, and set a geographic orientation deviation threshold; for any consumption record, if the deviation between the consumption location of the consumption record at the time of consumption and the user's actual geographic location is greater than the geographic orientation deviation threshold, the current consumption record is determined to lack authenticity and is an abnormal reimbursement record, and authenticity anomaly feedback is sent to the user and the reimbursement reviewer; if the deviation between the consumption location of the consumption record at the time of consumption and the user's actual geographic location is less than or equal to the geographic orientation deviation threshold, the current consumption record is determined to have passed the authenticity verification;

[0072] In specific implementation, the cloud platform monitors the time when consumption records are generated and the actual geographical location of the user in real time, verifies whether the deviation between the actual geographical location of the user at the time when the consumption records are generated and the geographical location when the consumption records are uploaded is within the threshold range, and makes authenticity judgments on the user's consumption records to avoid false consumption and reimbursement fraud;

[0073] Determine that consumption records that meet the reimbursement standards and pass authenticity verification are reimbursable records, monitor the user's consumption records when on a business trip in real time through the cloud platform and screen out reimbursable records, use the characteristic parameter identification method to collect the current reimbursable record characteristic parameters, and then predict the abnormal probability of the current reimbursable record based on the travel reimbursement abnormal probability model.

[0074] Step S400 includes the following contents:

[0075] Monitor and collect the time intervals of all reimbursable records that have passed authenticity verification, and then calculate the overlap between reimbursable records a and b according to the formula:

[0076]

[0077] Among them, C a,b is the overlap between reimbursable records a and b, Δt is the consumption time interval between reimbursable records a and b, exp() is the exponential function of natural logarithm, m is the attribute characteristic parameter number of the reimbursable record, n is the numerical characteristic parameter number of the reimbursable record, and n m is the number of attribute characteristic parameters of reimbursable records, n n is the number of numerical characteristic parameters of reimbursable records, x a,m is the parameter value of the attribute characteristic parameter m of the reimbursable record a, x b,m is the parameter value of the attribute characteristic parameter m of the reimbursable record b, x a,n is the parameter value of the numerical characteristic parameter n of the reimbursable record a, x b,n is the parameter value of the numerical characteristic parameter n of the reimbursable record b, k m and k n are the evaluation weights of the attribute characteristic parameter m and the numerical characteristic parameter n of the reimbursable record, respectively; |x| is the absolute value of x; flag() is the flag function, when the condition in the brackets is met, the function output is 1, when the condition in the brackets is not met, the function output is 0; max() is the maximum value function, the function output value is the maximum value of all the data in the brackets;

[0078] In specific implementation, the consumption time period can be collected to calculate the time interval. When the consumption time period of any consumption record overlaps with the current catering consumption record time period, the time interval between the two consumption records is judged to be negative, and the overlap degree of the two consumption records is corrected to 1; according to the overlap calculation formula, the result value range is (0,1], so when the consumption time periods of the two consumption records overlap, the overlap degree of the two consumption records is corrected to the maximum value;

[0079] For the above calculation formula, the following calculation example is given:

[0080] Assume that both reimbursable records a and b have passed compliance verification and authenticity verification and are reimbursable records; among them, the parameters of a are as follows: consumption type: transportation, consumption location: Beijing, consumption amount: 100; the parameters of b are as follows: consumption type: transportation, consumption location: Beijing, consumption amount: 120; consumption type calculation weight: 0.6, consumption location calculation weight: 0.4, consumption amount calculation weight: 0.7, time interval between reimbursable records a and b: 2, calculation unit: hour; it can be calculated that the overlap of reimbursable records a and b is 12.58%.

[0081] Step S500 includes the following contents:

[0082] The reimbursable records are divided into automatic reimbursement records, manual review records, and rejected reimbursement records;

[0083] Calculate the overlap between the user's reimbursable records and set the overlap threshold. For any reimbursable record, if there is another reimbursable record with an overlap greater than or equal to the overlap threshold, it is determined that there is a risk of duplicate reimbursement between the two reimbursable records, and both reimbursable records are classified as manual review records; if the overlap between the current reimbursable record and all other reimbursable records is less than the overlap threshold, it is determined that there is no risk of duplicate reimbursement for the current reimbursable record, and subsequent review and analysis of the current reimbursable record is performed;

[0084] In specific implementation, since the overlap between consumption records with overlapping consumption time periods is corrected to 1, when determining the overlap, consumption records with overlapping consumption time periods will be judged as having a risk of duplicate reimbursement and will be classified as manual review records, thereby avoiding the risk of duplicate reimbursement for users;

[0085] The abnormal probability P of the current reimbursable record is analyzed by the travel reimbursement abnormal probability model, and the manual review boundary [P1, P2] is set; when P<P1, the current reimbursable record is classified as an automatic reimbursement record; when P1≤P≤P2, the current reimbursable record is classified as a manual review record; when P>P2, the current reimbursable record is classified as a rejected reimbursement record;

[0086] For automatic reimbursement records, the cloud platform will collect statistics on all automatic reimbursement records of users during business trips and automatically reimburse them through the system at the end of the business trip; for rejected reimbursement records, the user will be sent a rejected reimbursement feedback and be reminded to submit paper review materials for reimbursement review at the end of the business trip;

[0087] For manual review records, calculate the review priority evaluation values ​​of all manual review records, and submit manual review applications in descending order according to the calculation results; for any manual review record a, according to the formula:

[0088]

[0089] Among them, L a is the audit priority evaluation value of manual audit record a, T is the audit waiting time of manual audit record a, k1 is the abnormal probability evaluation coefficient of manual audit record a, P a is the abnormal probability of manual review record a, k2 is the duplicate reimbursement evaluation coefficient of manual review record a, j is the reimbursable record number whose overlap with manual review record a is greater than or equal to the overlap threshold, n j is the number of reimbursable records whose overlap with the manually reviewed record a is greater than or equal to the overlap threshold, C a，j The overlap between the manually reviewed record a and the reimbursable record j;

[0090] For the above calculation formula, the following calculation example is given: Assume that the review waiting time for manual review record a is 5, the calculation time is days, the abnormal probability assessment coefficient is 0.8, the duplicate reimbursement assessment coefficient is 0.5, the abnormal probability of manual review record a is 0.7, and the overlap threshold is set to 0.1. There are three other reimbursable records with overlaps of 0.15, 0.08, and 0.12, respectively. Among them, the value with an overlap of 0.08 with manual review record a is less than the overlap threshold, so it is not included in the calculation. The manual review priority assessment value of manual review record a is calculated to be 3.475.

[0091] like Figure 2 As shown, the present invention also provides a travel reimbursement intelligent accounting management system based on a cloud platform, including a feature processing module, an abnormality analysis module, and a reimbursement processing module;

[0092] The feature processing module is used to collect the feature parameters of each consumable record, and convert the attribute parameters into numerical parameters through encoding processing to train the travel reimbursement abnormality probability model; the abnormality analysis module is used to monitor and collect the consumption records of users when they are on business trips in real time, and screen the users' business trip consumption records, and analyze the abnormal probability and overlap of each reimbursable record; the reimbursement processing module classifies the various reimbursable records into types through threshold judgment, and then performs reimbursement processing and feedback separately, and analyzes the audit priority evaluation value of the manually reviewed records.

[0093] The feature processing module includes a feature parameter acquisition unit, an attribute encoding unit, and a model data set construction unit;

[0094] The feature parameter collection unit is used to screen and identify all feature parameters that affect the judgment of travel reimbursement anomalies; the attribute encoding unit is used to convert the attribute parameters in the screened feature parameters into numerical parameters using one-hot encoding to participate in subsequent analysis; the model data set construction unit is used to combine all feature parameters with numerical data and corresponding travel reimbursement anomaly judgment results to construct a travel reimbursement data set.

[0095] The abnormal analysis module includes a user consumption record collection unit, a compliance verification unit, an authenticity verification unit, an abnormal reimbursement abnormal probability analysis unit, and a coincidence calculation unit;

[0096] The user consumption record collection unit is used to monitor and collect various consumption records of users during business trips in real time; the compliance verification unit is used to verify whether the user's consumption records comply with the company's reimbursement standards; the authenticity verification unit is used to analyze the consumption time and place of the user's consumption records and verify the authenticity of the consumption records; the abnormal reimbursement abnormal probability analysis unit uses the reimbursement abnormal probability model to analyze the abnormal probability of each reimbursable record; the overlap calculation unit is used to calculate the overlap data between each reimbursable record.

[0097] The reimbursement processing module includes a reimbursement division unit, an automatic reimbursement unit, a manual review unit, and a reimbursement rejection unit;

[0098] The reimbursement classification unit classifies various reimbursable records by threshold judgment; the automatic reimbursement unit counts all automatic reimbursement records of users during business trips through the cloud platform and automatically reimburses through the system at the end of the business trip; the manual review unit is used to analyze the review priority of manual review records and submit manual reimbursement reviews in order from high to low priority; the reimbursement rejection unit sends reimbursement rejection feedback to the user and reminds the user to submit paper review materials for reimbursement review at the end of the business trip.

[0099] 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. A cloud platform-based intelligent accounting management method for travel reimbursement, characterized in that , the method comprises the following analysis steps: Step S100: Acquire travel reimbursement history data and reimbursement review results, screen and identify feature parameters of all historical data, perform type conversion on the feature parameters and construct a data set; Step S200: construct a logistic regression model and use the data set to train the model to obtain a travel reimbursement abnormality probability model under different feature parameters; Step S300: obtaining the consumption records of the user during the travel process through the cloud platform, determining whether each record can be reimbursed, collecting characteristic parameters of the reimbursable records, and using the travel reimbursement abnormal probability model to analyze the abnormal probability of each reimbursable record; Step S400: Calculate and analyze the overlap of various reimbursable records based on the characteristic parameters of the reimbursable records; Step S500: Perform threshold judgment on the abnormal probability and overlap data of the reimbursable records, classify the processing types of the reimbursable records, and process different types of reimbursable records separately.

2. According to claim 1, a cloud platform-based intelligent accounting management method for travel reimbursement is characterized in that The step S100 includes the following contents: Collect historical data on travel reimbursement, filter and identify all characteristic parameters that affect the judgment of travel reimbursement anomalies; the characteristic parameters are divided into attribute characteristic parameters and numerical characteristic parameters; For attribute-type feature parameters, use one-hot encoding to convert each attribute type in the attribute-type feature parameter into a numeric data format; for any attribute-type feature parameter, analyze all possible subtypes in the current feature parameter, set a binary code string with a length equal to the number of all subtypes, and each bit in the code string corresponds to a different subtype; for any subtype in the current feature parameter, set the value of the corresponding bit in the code string to 1, and the remaining bits to 0, thereby converting all subtypes in the current feature parameter into different numeric data; All feature parameters are combined with numerical data and the corresponding travel reimbursement anomaly judgment results to construct a travel reimbursement dataset.

3. According to claim 2, a cloud platform-based intelligent accounting management method for travel reimbursement is characterized in that The step S200 includes the following contents: A travel reimbursement abnormal probability model is constructed and the travel reimbursement dataset is used for model training. The model is as follows: Among them, P(Rei Unusual |X) is the abnormal probability of travel reimbursement predicted using the feature parameter combination X, exp() is the exponential function of the natural logarithm, i is the feature parameter number in the feature parameter combination X, n X is the number of characteristic parameters in the characteristic parameter combination X, α0 is the constant parameter of the travel reimbursement abnormal probability model, α i is the weight coefficient of each feature parameter in the feature parameter combination X in the travel reimbursement abnormal probability model, X i is the value of feature parameter i of travel reimbursement data in feature parameter combination X.

4. According to claim 2, a cloud platform-based intelligent accounting management method for travel reimbursement is characterized in that The step S300 includes the following contents: According to the business trip reimbursement standards set by the enterprise, the range of various characteristic parameters of the consumption records of the user during the business trip is limited. When any consumption record of the user during the business trip meets the limited range of all characteristic parameters, the current consumption record is determined to meet the reimbursement standards; Collect the consumption location and the user's actual geographic location at the time of consumption for each consumption record, and set a geographic deviation threshold; for any consumption record, if the deviation between the consumption location of the consumption record at the time of consumption and the user's actual geographic location is greater than the geographic deviation threshold, determine that the current consumption record lacks authenticity and is an abnormal reimbursement record, and send authenticity abnormality feedback to the user and the reimbursement reviewer; If the deviation between the consumption location of the consumption record and the user's actual geographical location at the consumption time is less than or equal to the geographical orientation deviation threshold, the current consumption record is determined to have passed the authenticity verification; Consume records that are determined to meet the reimbursement criteria and pass the authenticity verification are regarded as reimbursable records. The consumption records of users during business trips are monitored in real time through the cloud platform, and the reimbursable records are screened out. The characteristic parameters of the current reimbursable records are collected using the described characteristic parameter identification method, and then the abnormal probability of the current reimbursable records is predicted according to the business trip reimbursement abnormal probability model.

5. According to claim 2, a cloud platform-based intelligent accounting management method for travel reimbursement is characterized in that The following content is included in step S400: Monitor and collect the consumption time intervals of all reimbursable records that pass the authenticity verification, and then calculate the coincidence degree of reimbursable records a and b according to the formula: Among them, C a，b is the overlap between reimbursable records a and b, Δt is the consumption time interval between reimbursable records a and b, exp() is the exponential function of natural logarithm, m is the attribute characteristic parameter number of the reimbursable record, n is the numerical characteristic parameter number of the reimbursable record, and n m is the number of attribute characteristic parameters of reimbursable records, n n is the number of numerical characteristic parameters of reimbursable records, x a，m is the parameter value of the attribute characteristic parameter m of the reimbursable record a, x b,m is the parameter value of the attribute characteristic parameter m of the reimbursable record b, x a，n is the parameter value of the numerical characteristic parameter n of the reimbursable record a, x b,n is the parameter value of the numerical characteristic parameter n of the reimbursable record b, k m and k n are the evaluation weights of the attribute characteristic parameter m and the numerical characteristic parameter n of the reimbursable record, respectively; |x| is the absolute value of x; flag() is the flag function, when the condition in the brackets is met, the function output is 1, when the condition in the brackets is not met, the function output is 0; max() is the maximum value function, and the function output value is the maximum value of all the data in the brackets.

6. According to claim 2, a cloud platform-based intelligent accounting management method for travel reimbursement is characterized in that The following content is included in step S500: Classify the reimbursable records into automatically reimbursed records, manually reviewed records, and rejected reimbursement records; Calculate the coincidence degree between the reimbursable records of the user, set the coincidence degree threshold. For any reimbursable record, if there is another reimbursable record whose coincidence degree with the current reimbursable record is greater than or equal to the coincidence degree threshold, it is determined that there is a risk of duplicate reimbursement between the two reimbursable records, and both reimbursable records are classified as manually reviewed records; if the coincidence degree between the current reimbursable record and all other reimbursable records is less than the coincidence degree threshold, it is determined that there is no risk of duplicate reimbursement for the current reimbursable record, and subsequent review and analysis are performed on the current reimbursable record; Analyze the abnormal probability P of the current reimbursable record through the business trip reimbursement abnormal probability model, and set the manual review boundary [P1, P2]; when P < P1, classify the current reimbursable record as an automatically reimbursed record; when P1 ≤ P ≤ P2, classify the current reimbursable record as a manually reviewed record; when P > P2, classify the current reimbursable record as a rejected reimbursement record; For automatically reimbursed records, count all the automatically reimbursed records of the user during the business trip through the cloud platform and perform automatic reimbursement through the system at the end of the business trip; for rejected reimbursement records, send a rejected reimbursement feedback to the user and remind the user to submit paper review materials for reimbursement review at the end of the business trip; For manually reviewed records, calculate the review priority evaluation values of all manually reviewed records and submit manual review applications in descending order according to the calculation results; For any manually reviewed record a, according to the formula: Among them, L a is the audit priority evaluation value of manual audit record a, T is the audit waiting time of manual audit record a, k1 is the abnormal probability evaluation coefficient of manual audit record a, P a is the abnormal probability of manually reviewed record a, k2 is the duplicate reimbursement evaluation coefficient of manually reviewed record a, j is the reimbursable record number whose overlap with manually reviewed record a is greater than or equal to the overlap threshold, n j is the number of reimbursable records whose overlap with the manually reviewed record a is greater than or equal to the overlap threshold, C a，j It is the overlap between the manually reviewed record a and the reimbursable record j.

7. A cloud-based intelligent travel reimbursement accounting management system using a cloud-based intelligent travel reimbursement accounting management method according to any one of claims 1 to 6, characterized in that: The system includes a feature processing module, an abnormal analysis module, and a reimbursement processing module; The feature processing module is used to collect the characteristic parameters of each consumable record, and convert the attribute type parameters into numerical type parameters through coding processing for training the business trip reimbursement abnormal probability model; the abnormal analysis module is used to monitor and collect the consumption records of users during business trips in real time, screen the consumption records of users during business trips, and analyze the abnormal probability and coincidence degree of each reimbursable record; the reimbursement processing module classifies each reimbursable record through threshold judgment and then performs reimbursement processing and feedback respectively, and analyzes the review priority evaluation value of the manually reviewed records.

8. According to claim 7, a cloud platform-based intelligent accounting and management system for travel reimbursement is characterized in that: The feature processing module includes a feature parameter collection unit, an attribute coding unit, and a model data set construction unit; The feature parameter collection unit is used to screen and identify all feature parameters that affect the judgment of travel reimbursement anomalies; the attribute encoding unit is used to convert the attribute parameters in the screened feature parameters into numerical parameters using one-hot encoding to participate in subsequent analysis; the model data set construction unit is used to combine all feature parameters with numerical data and corresponding travel reimbursement anomaly judgment results to construct a travel reimbursement data set.

9. According to claim 7, a cloud platform-based intelligent travel reimbursement accounting management system is characterized in that: The abnormal analysis module includes a user consumption record collection unit, a compliance verification unit, an authenticity verification unit, an abnormal reimbursement abnormal probability analysis unit, and a coincidence calculation unit; The user consumption record collection unit is used to monitor and collect various consumption records of users during business trips in real time; the compliance verification unit is used to verify whether the various consumption records of users comply with the company's reimbursement standards; the authenticity verification unit is used to analyze the consumption time and place of various consumption records of users to verify the authenticity of various consumption records; the abnormal reimbursement abnormal probability analysis unit uses the reimbursement abnormal probability model to analyze the abnormal probability of various reimbursable records; the overlap calculation unit is used to calculate the overlap data between various reimbursable records.

10. The cloud platform-based intelligent travel reimbursement accounting management system according to claim 7, characterized in that: The reimbursement processing module includes a reimbursement division unit, an automatic reimbursement unit, a manual review unit, and a reimbursement rejection unit; The reimbursement classification unit classifies various reimbursable records into types through threshold judgment; the automatic reimbursement unit counts all automatic reimbursement records of the user during the business trip through the cloud platform and automatically reimburses through the system at the end of the business trip; the manual review unit is used to analyze the review priority of the manual review records and submit manual reimbursement reviews in order from high to low priority; the reimbursement refusal unit sends reimbursement refusal feedback to the user and reminds the user to submit paper review materials for reimbursement review at the end of the business trip.

Images