Reimbursement approval method and related product
By extracting keywords and identifying business scenarios from expense reimbursement data, and automating the approval process, the system solves the problems of low efficiency and poor accuracy in traditional methods, achieving efficient and precise financial management.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 太保科技有限公司
- Filing Date
- 2026-01-07
- Publication Date
- 2026-04-17
AI Technical Summary
Traditional manual approval methods and existing intelligent solutions are insufficient to meet the needs of enterprises for efficient, accurate and flexible financial management, especially in the process of end-to-end expense reimbursement approval, where there are problems of low efficiency and poor accuracy.
By acquiring target expense reimbursement data, extracting keywords, determining target business scenarios and scenario weights, determining approval rules based on scenarios and weights, automating the approval process, and generating exception reports and visualization reports.
It improved the accuracy and efficiency of reimbursement approval, reduced manual intervention, standardized approval criteria, and ensured the fairness and standardization of approval results.
Smart Images
Figure CN121883192A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology, and in particular to a reimbursement approval method and related products. Background Technology
[0002] In the daily operations of a company, the expense reimbursement approval process is a crucial part of financial management, as it directly relates to the effective use of company funds and risk control.
[0003] However, as businesses continue to expand and become more complex, traditional manual approval methods and existing intelligent solutions are struggling to meet the needs of efficient, accurate, and flexible financial management when dealing with the entire expense reimbursement approval process. Summary of the Invention
[0004] In view of the above problems, this application provides a reimbursement approval method and related products, aiming to improve the accuracy and processing efficiency of reimbursement approval.
[0005] The embodiments of this application disclose the following technical solutions:
[0006] In a first aspect, embodiments of this application provide a reimbursement approval method, the method comprising:
[0007] Obtain the target expense reimbursement data to be processed; the target expense reimbursement data includes the target expense reimbursement text, target scenario tags, and target project information;
[0008] Keyword extraction is performed on the target expense reimbursement text to obtain the target keywords;
[0009] The target scene library is determined from the preset scene database based on the target scene tags;
[0010] Based on the target scenario library, the target business scenario and target scenario weight corresponding to the target reimbursement text are determined according to the target keywords and target project information;
[0011] Determine the target approval rules based on the target business scenario and the target scenario weight;
[0012] The target reimbursement data is approved according to the target approval rules to obtain the target approval result corresponding to the target reimbursement data.
[0013] Optionally, target approval rules are determined based on the target business scenario and the target scenario weight, including:
[0014] Determine n rules based on the target business scenario; n is an integer greater than or equal to 1.
[0015] Sort the n rules according to the target scenario weight;
[0016] The target approval rule is determined based on the sorted n rules.
[0017] Optionally, if the target approval result is characterized as abnormal, it also includes:
[0018] Retrieve historical expense reimbursement data associated with the target expense reimbursement data;
[0019] Identify abnormal characteristics of the target based on target reimbursement data and historical reimbursement data;
[0020] Determine the alarm level of the target expense reimbursement data based on the target's abnormal characteristics;
[0021] Determine the anomaly tracing map corresponding to the target expense reimbursement data based on the target expense reimbursement data;
[0022] A target anomaly report is generated based on the alarm level and the anomaly source map.
[0023] Optionally, the method further includes:
[0024] Retrieve k expense reimbursement data within a preset time period; where k is an integer greater than or equal to 1;
[0025] Calculate the compliance rate and the percentage of abnormal types based on k expense reimbursement data;
[0026] Generate a target statistics table based on compliance rate and the proportion of anomaly types;
[0027] Analyze the target statistics table to obtain the target analysis results;
[0028] The target visualization report is determined based on the target statistics table and target analysis results.
[0029] Secondly, embodiments of this application provide a reimbursement approval device, which includes:
[0030] The acquisition unit is used to acquire the target reimbursement data to be processed; wherein, the target reimbursement data includes the target reimbursement text, the target scenario tag, and the target project information;
[0031] The extraction unit is used to extract keywords from the target reimbursement text to obtain the target keywords.
[0032] The determining unit is used to determine the target scene library from a preset scene database based on the target scene label;
[0033] The scenario determination unit is used to determine the target business scenario and target scenario weight corresponding to the target reimbursement text based on the target scenario library, target keywords and target project information;
[0034] The approval rule determination unit is used to determine the target approval rule based on the target business scenario and the target scenario weight.
[0035] The approval unit is used to approve the target reimbursement data according to the target approval rules and obtain the target approval result corresponding to the target reimbursement data.
[0036] Optionally, the approval rule determination unit is used for:
[0037] Determine n rules based on the target business scenario; n is an integer greater than or equal to 1.
[0038] Sort the n rules according to the target scenario weight;
[0039] The target approval rule is determined based on the sorted n rules.
[0040] Optionally, if the target approval result is characterized as abnormal, the system further includes: an abnormality report generation unit, used for:
[0041] Retrieve historical expense reimbursement data associated with the target expense reimbursement data;
[0042] Identify abnormal characteristics of the target based on target reimbursement data and historical reimbursement data;
[0043] Determine the alarm level of the target expense reimbursement data based on the target's abnormal characteristics;
[0044] Determine the anomaly tracing map corresponding to the target expense reimbursement data based on the target expense reimbursement data;
[0045] A target anomaly report is generated based on the alarm level and the anomaly source map.
[0046] Optionally, the device further includes: a visualization report generation unit, for:
[0047] Retrieve k expense reimbursement data within a preset time period; where k is an integer greater than or equal to 1;
[0048] Calculate the compliance rate and the percentage of abnormal types based on k expense reimbursement data;
[0049] Generate a target statistics table based on compliance rate and the proportion of anomaly types;
[0050] Analyze the target statistics table to obtain the target analysis results;
[0051] The target visualization report is determined based on the target statistics table and target analysis results.
[0052] Thirdly, embodiments of this application provide a control device, including a processor and a memory, wherein the memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to complete the reimbursement approval method as described in the first aspect.
[0053] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which is loaded by a processor to execute the reimbursement approval method as described in the first aspect.
[0054] Beneficial effects:
[0055] The reimbursement approval method provided in this application involves acquiring the target reimbursement data to be processed, extracting keywords from the target reimbursement text to obtain target keywords, determining a target scenario library from a preset scenario database based on target scenario tags, further determining the target business scenario and target scenario weight corresponding to the target reimbursement text based on the target scenario library, target keywords, and target project information, determining target approval rules based on the target business scenario and target scenario weight, and finally approving the target reimbursement data according to the target approval rules to obtain the target approval result corresponding to the target reimbursement data. The target reimbursement data includes the target reimbursement text, target scenario tags, and target project information.
[0056] In this way, by extracting keywords from the target expense reimbursement text and determining the target business scenario based on the obtained target keywords and target project information, it is possible to determine the exclusive business scenario to which the target expense reimbursement data belongs. Then, by combining the target business scenario and target weight, the target approval rules for approving the target expense reimbursement data can be determined, thereby improving the accuracy of expense reimbursement approval. At the same time, this application improves the processing efficiency of expense reimbursement approval by automating the approval process without human intervention. Attached Figure Description
[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0058] Figure 1 A flowchart illustrating a reimbursement approval method provided in this application embodiment;
[0059] Figure 2 A schematic diagram of the structure of a reimbursement approval device provided in this application embodiment;
[0060] Figure 3 This is a schematic diagram of the structure of a control device provided in an embodiment of this application. Detailed Implementation
[0061] As described earlier, in the traditional manual approval model, basic approval work is highly repetitive. Approving personnel need to spend a lot of time manually verifying document information and comparing each rule one by one. This process is not only tedious and monotonous, but also prone to errors due to human negligence, thus affecting approval efficiency and accuracy.
[0062] Based on this, embodiments of this application provide a reimbursement approval method and related products. The method includes: acquiring target reimbursement data to be processed; extracting keywords from the target reimbursement text to obtain target keywords; determining a target scenario library from a preset scenario database based on target scenario tags; further determining the target business scenario and target scenario weight corresponding to the target reimbursement text based on the target scenario library, target keywords, and target project information; determining target approval rules based on the target business scenario and target scenario weight; and finally, approving the target reimbursement data according to the target approval rules to obtain the target approval result corresponding to the target reimbursement data. The target reimbursement data includes the target reimbursement text, target scenario tags, and target project information.
[0063] In this way, by extracting keywords from the target expense reimbursement text and determining the target business scenario based on the obtained target keywords and target project information, it is possible to determine the exclusive business scenario to which the target expense reimbursement data belongs. Then, by combining the target business scenario and target weight, the target approval rules for approving the target expense reimbursement data can be determined, thereby improving the accuracy of expense reimbursement approval. At the same time, this application improves the processing efficiency of expense reimbursement approval by automating the approval process without human intervention.
[0064] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.
[0065] The collection and processing of relevant data (including but not limited to experimental data, test data, simulation data, user data, etc.) involved in this application shall strictly comply with the requirements of national laws and regulations when applied in the following embodiments, obtain the informed consent or separate consent of the subject obtaining the data information, and carry out data use and processing within the scope of laws and regulations and the authorization of the subject.
[0066] See Figure 1 The figure is a flowchart illustrating a reimbursement approval method provided in an embodiment of this application.
[0067] It should be noted that the embodiments of this application can be implemented based on a "client-server" architecture, wherein the server side adds a scene processing module, an early warning and source tracing module, a knowledge graph module and a data review module, while optimizing the intelligent review module and the AI recognition module.
[0068] Combination Figure 1 As shown in the embodiments of this application, the reimbursement approval method may include:
[0069] S11: Obtain the target expense reimbursement data to be processed.
[0070] Target reimbursement data refers to data used for reimbursement. In one possible implementation, target reimbursement data can be various types of data, such as images (e.g., photos of paper invoices), text (e.g., text content of electronic reimbursement forms), and numbers (e.g., invoice amount figures). Specifically, target reimbursement data can be image documents, electronic document documents, or electronic ledger data of invoices, etc., without specific limitations here.
[0071] In this embodiment of the application, the target reimbursement data includes the target reimbursement text, the target scenario tag, and the target project information.
[0072] The target reimbursement text refers to text-formatted data used for reimbursement. It should be understood that, in this embodiment, if the input target reimbursement data is not in text format, the target reimbursement data can be processed to obtain the target reimbursement text.
[0073] As an example, taking the AI recognition module as an example, the AI recognition module has the ability to cross-validate multimodal data. The input of the AI recognition module can be image documents, electronic document documents, electronic invoice ledger data, etc.; the output of the AI recognition module can be structured information after cross-validation.
[0074] The processing flow of the AI recognition module can be as follows:
[0075] Step 11: Recognize the acquired image to be processed, and call the electronic ledger data to verify the authenticity of the recognition results (such as invoice code and amount).
[0076] Step 12: Cross-validate the "project name" extracted from the electronic document with the project information in the knowledge graph.
[0077] Step 13: Mark information with an error greater than 5% (such as the difference between handwritten amount and OCR recognition) as "high risk to be verified" to improve recognition accuracy.
[0078] It should be understood that, in the embodiments of this application, cross-validation of multimodal data improves the accuracy and reliability of data identification and reduces the workload of manual verification. Simultaneously, marking high-risk information can remind approvers to pay close attention, thereby reducing financial risks.
[0079] S12: Extract keywords from the target reimbursement text to obtain target keywords.
[0080] Keyword extraction processing refers to extracting key information from the target expense report. The target keywords can summarize the core content of the expense report, such as the type of expense report (travel, office purchase, etc.), the items involved (airfare, office supplies, etc.), and the amount of expense.
[0081] It should be understood that condensing a large amount of text information into key keywords facilitates subsequent processing and analysis, and improves data processing efficiency; at the same time, the extracted keywords can directly reflect the key information of the reimbursement matter, which helps to quickly locate and understand the core content of the reimbursement.
[0082] S13: Determine the target scene library from the preset scene database based on the target scene label.
[0083] The target scenario tag refers to the unique identifier of the scenario to which the target reimbursement data belongs.
[0084] The preset scene library refers to a collection of multiple scene libraries, and the scene library refers to the collection of all scenes corresponding to a vertical domain.
[0085] It should be understood that the preset scenario database stores information related to various business scenarios. The target scenario tag serves as an index, which can accurately locate the target scenario library that matches the current reimbursement item from the database. This scenario library contains detailed rules and feature information under specific business scenarios.
[0086] It should be understood that using scene tags to quickly and accurately find the corresponding scene library avoids blindly searching through large amounts of data, improving data processing speed and accuracy; at the same time, managing different business scenarios separately makes it easier to formulate specific rules and processes for different scenarios, improving the level of management refinement.
[0087] S14: Based on the target scenario library, determine the target business scenario and target scenario weight corresponding to the target reimbursement text according to the target keywords and target project information.
[0088] The target business scenario refers to the actual business scenario to which the target expense reimbursement data belongs.
[0089] The target scenario weight refers to the importance of the target business scenario.
[0090] It should be understood that, in this embodiment of the application, based on the target scenario library, and combined with extracted target keywords and target project information, the specific business scenario to which the reimbursement item belongs is further clarified, and the importance of the scenario in the overall reimbursement is determined, i.e., the target scenario weight. In this way, by comprehensively considering multi-dimensional information to determine the business scenario, the limitations of single-factor judgment are avoided, and the accuracy of business scenario identification is improved. At the same time, quantifying the importance of different business scenarios through scenario weight provides a basis for the formulation of subsequent approval rules.
[0091] In one possible implementation, a scenario recognition module can be deployed on the server side. The input to the scenario recognition module is the scenario tag submitted by the person seeking reimbursement (i.e., the target scenario tag), the text of the reason for reimbursement (i.e., the target reimbursement text), and the associated project / customer information (i.e., the target project information). The output of the scenario recognition module can be the matched business scenario (i.e., the target business scenario) and the scenario weight (i.e., the target weight) (e.g., "cross-regional customer visit" weight 95%).
[0092] The processing flow of the scene recognition module can be as follows:
[0093] 1) Based on the BERT semantic model, parse the expense report text and extract target keywords (such as "customer visit" and "cross-regional").
[0094] BERT semantic model is a pre-trained language representation model that can understand the semantic information of text and can be used for text classification, information extraction and other tasks in natural language processing.
[0095] As an example, suppose the target expense report is a report on the reason for the expense (e.g., "I went to Shanghai in October 2023 to attend Client A's annual product launch, and I am applying for travel expenses"). Then, a pre-trained BERT model can be used to perform semantic analysis, extract keywords and contextual relationships, and output structured keywords: ["Shanghai", "Client A", "Annual Product Launch", "Travel Expenses"].
[0096] It should be understood that the BERT model can capture long-distance dependencies through a bidirectional Transformer encoder, thus resolving the problem of polysemy (e.g., "Customer A" is a name rather than a common noun). Furthermore, domain-specific fine-tuning (e.g., incorporating expense reimbursement corpora) can improve the accuracy of specialized terminology recognition.
[0097] 2) Determine the project type by associating with the project database, and determine the customer level by associating with the customer database.
[0098] As an example, you can input target project information, including project ID, customer ID, etc. Furthermore, you can retrieve project type (e.g., "marketing activity" or "R&D"), budget item, etc., from the project database based on the project ID. Simultaneously, you can retrieve customer level (e.g., "VIP customer" or "regular customer"), cooperation duration, etc., from the customer database based on the customer ID. For example, the "project type" is "marketing activity" and the "customer level" is "VIP customer".
[0099] 3) Match business scenarios through the scenario rule engine (preset scenario keyword library) and calculate the matching weight (≥80% is judged as a valid scenario).
[0100] A scenario rule engine refers to an engine used to match and process input information according to preset rules. Here, it is used to match business scenarios and calculate weights.
[0101] As an example, a scenario rule base design could be:
[0102] Rules are defined in the form of combinations of conditions, for example:
[0103] Rule 1: IF Project Type="Marketing Event" AND Keyword Contains "Press Conference" AND Client Level="VIP Client" THEN Scenario="VIP Client Marketing Event Travel"
[0104] Rule 2: IF Project Type="R&D" AND Keyword Contains "Testing" THEN Scenario="R&D Project Testing Travel"
[0105] Weight calculation: When each rule is successfully matched, a weight is assigned according to the number and importance of the conditions (e.g., rule 1 weight = 0.8, rule 2 weight = 0.6).
[0106] Correspondingly, the matching process for business scenarios can be as follows: traverse the rule base, calculate the matching degree between the current expense report and each rule; select the rule with the highest matching degree as the primary scenario, and merge other high-weight rules as secondary scenarios.
[0107] As an example, the "primary scenario" is "VIP customer marketing activity travel" with a "weight" of 0.8; the "secondary scenario" is "intercity travel" with a "weight" of 0.3.
[0108] 4) Push the scene information to the intelligent review module, call the corresponding scene rule library, and achieve accurate adaptation of special scene rules.
[0109] S15: Determine the target approval rules based on the target business scenario and the target scenario weight.
[0110] Target approval rules refer to the set of rules used to approve target expense reimbursement data.
[0111] In one possible implementation, step S15 may include:
[0112] A1: Determine n rules based on the target business scenario.
[0113] Where n is an integer greater than or equal to 1.
[0114] As an example, the target business scenario can correspond to general rules and scenario rules.
[0115] A2: Sort the n rules according to the target scenario weight.
[0116] As an example, if the target scene weight is greater than the preset weight threshold, the scene rule takes precedence over the general rule; if the target scene weight is less than or equal to the preset weight threshold, the general rule takes precedence over the scene rule.
[0117] A3: Determine the target approval rule based on the sorted n rules.
[0118] It should be understood that the embodiments of this application flexibly adjust the approval rules according to different business scenarios and weights to ensure that the approval rules match actual business needs and improve the rationality and effectiveness of the approval process. At the same time, the order of application of different rules is clarified by rule sorting, making the approval process more organized and avoiding rule conflicts and confusion.
[0119] As an example, the server can add a scenario rule adaptation engine to the intelligent review module. The input to the intelligent review module is scenario information, a basic rule base, and a scenario rule base; the output of the intelligent review module is the review result after scenario adaptation. The process for determining the target approval rule in the intelligent review module can be as follows:
[0120] 1) The basic rule base consists of general rules (such as the standard for ordinary business travel and accommodation), while the scenario rule base consists of specific rules (such as the standard for customer reception scenarios).
[0121] The basic rule base refers to a database containing general audit rules, such as standard travel and accommodation rules.
[0122] A scenario rule base refers to a database containing specific audit rules for particular business scenarios, such as customer reception scenario standards.
[0123] 2) If a valid scenario is identified, the principle of "scenario rules take precedence, basic rules provide a safety net" shall be adopted (e.g., the catering standards for customer reception scenarios cover general standards).
[0124] 3) In case of rule conflict (such as conflict between scenario rules and job level rules), the default priority shall be used to determine the priority (business scenario priority > job level priority).
[0125] In this way, the intelligent review module can flexibly adapt review rules according to different business scenarios, improving the relevance and accuracy of the review. At the same time, the preset priority judgment of rule conflicts ensures the rationality and consistency of the review results.
[0126] S16: Approve the target reimbursement data according to the target approval rules and obtain the target approval result corresponding to the target reimbursement data.
[0127] It should be understood that, in this embodiment of the application, the collected target reimbursement data is comprehensively reviewed in accordance with the determined target approval rules to determine whether the reimbursement items comply with the regulations, and finally the target approval result such as approval or abnormality is given.
[0128] Thus, by automatically approving according to the target approval rules, approval efficiency is improved, and the time for manual approval and the impact of subjective factors are reduced. At the same time, the implementation of this application unifies the approval standards, ensuring that reimbursement matters under the same business scenario receive consistent approval results, thereby improving the fairness and standardization of the approval process.
[0129] In one possible implementation of the reimbursement approval method provided in the above embodiments, if the target approval result is characterized as abnormal, the method further includes:
[0130] B1: Obtain historical reimbursement data associated with the target reimbursement data.
[0131] Historical expense reimbursement data refers to historically approved data related to the target expense reimbursement data. For example, collecting historical expense reimbursement records related to current abnormal expense reimbursement data, including historical expense reimbursement information for the same person, the same project, or the same business scenario.
[0132] B2: Determine the abnormal characteristics of the target based on the target reimbursement data and historical reimbursement data.
[0133] It should be understood that, in the embodiments of this application, current abnormal reimbursement data and historical data can be compared and analyzed to identify the characteristics of target anomalies, such as abnormal fluctuations in reimbursement amount or abnormal reimbursement frequency.
[0134] B3: Determine the alarm level of the target reimbursement data based on the target's abnormal characteristics.
[0135] It should be understood that the embodiments of this application can classify different alarm levels, such as minor, general, and severe, according to the severity of the target abnormality characteristics, so as to take different countermeasures.
[0136] B4: Determine the anomaly tracing map corresponding to the target reimbursement data based on the target reimbursement data.
[0137] An anomaly attribution mapping refers to a structured graph generated by tracing the origins of anomalies in target expense reimbursement data. It should be understood that by analyzing the causes and processes that generate abnormal expense reimbursement data, an attribution mapping is constructed to identify the source of the problem and the involved stages.
[0138] B5: Generate a target anomaly report based on the alarm level and anomaly source map.
[0139] It should be understood that by integrating alarm levels and source tracing information to generate detailed reports, management personnel can be provided with a basis for decision-making.
[0140] In this embodiment, by acquiring historical data and conducting multi-faceted analysis, a comprehensive and in-depth understanding of abnormal situations is achieved, enabling not only to identify the abnormal manifestations but also to pinpoint the root causes. Simultaneously, alarm levels are determined based on abnormal characteristics, enabling hierarchical management, rational allocation of management resources, and priority handling of severe anomalies.
[0141] As an example, the server can deploy an early warning and source tracing module. The input of the early warning and source tracing module is abnormal data, historical reimbursement data, and related personnel data from the intelligent audit module. The output of the early warning and source tracing module is the abnormal warning level (high / medium / low) and the source tracing map.
[0142] The processing flow of the early warning and source tracing module can be as follows:
[0143] 1) Construct a multi-dimensional anomaly detection model. Input features include “amount fluctuation range (average of the past 3 months ± 50%)”, “frequency anomaly (more than 5 times of the same type of expense per month)”, “related reimbursements (more than 3 people reimbursing expenses in the same period and destination)”, and “high standard frequency (the proportion of low frequency high standard exceeds 30%)”.
[0144] 2) The model calculates the anomaly score x using the logistic regression algorithm. A score of x ≥ 80 indicates a high warning, a score of 60 > x > 80 indicates a medium warning, and a score of x < 60 indicates a low warning.
[0145] Logistic regression is a statistical method used for classification and prediction. It predicts the probability of an event occurring by establishing a logistic function, which is used here to calculate anomaly scores.
[0146] 3) Generate a traceability map and associate the abnormal data with the corresponding document information, historical records, and related personnel information (such as "Employee A's high-standard expense reimbursement this time - first occurrence - related customer is a strategic customer - no related expense reimbursement").
[0147] A source map refers to a graphical representation of relevant information about abnormal data, including document information, historical records, and information about related personnel, to facilitate tracing the cause of the anomaly.
[0148] 4) Synchronize the early warning information and source tracing map to the approval reference report to assist manual verification of anomalies.
[0149] It should be understood that multi-dimensional anomaly detection models can more comprehensively identify abnormal situations and improve the accuracy of anomaly identification. At the same time, the generated source map can intuitively display the source of abnormal data and related background information, assisting manual verification of anomalies and improving approval efficiency.
[0150] In one possible implementation of the reimbursement approval method provided in the above embodiments, it further includes:
[0151] C1: Retrieve k expense reimbursement data within a preset time period.
[0152] Where k is an integer greater than or equal to 1.
[0153] The preset time period refers to the time range within which expense reimbursement data needs to be statistically analyzed, such as a week, a month, a quarter, etc., without any specific limitation.
[0154] It should be understood that, in the embodiments of this application, multiple reimbursement data samples within a specific time period can be collected for statistical analysis.
[0155] C2: Calculate the compliance rate and the percentage of abnormal types based on k expense reimbursement data.
[0156] Compliance rate refers to the proportion of compliant expense reports out of the total number of reports.
[0157] The percentage of different anomaly types refers to the proportion of each type of anomaly among all anomalies.
[0158] It should be understood that, in the embodiments of this application, the proportion of reimbursement data that complies with the approval rules (compliance rate) and the proportion of different types of abnormal reimbursement data in the total can be statistically analyzed.
[0159] C3: Generate a target statistics table based on compliance rate and percentage of anomaly types.
[0160] A target statistics table refers to a statistical table that presents the calculation results in tabular form, clearly showing the overall situation and abnormal distribution of reimbursement data.
[0161] C4: Analyze the target statistics table to obtain the target analysis results.
[0162] It should be understood that, in the embodiments of this application, potential problems and trends are explored through in-depth analysis of statistical table data, such as which business scenarios have low compliance rates and which types of anomalies occur frequently.
[0163] C5: Determine the target visualization report based on the target statistics table and target analysis results.
[0164] It should be understood that, in the embodiments of this application, the target statistical table and analysis results can be displayed in the form of target visualization reports (such as bar charts, pie charts, line charts, etc.) to make the data more intuitive and easy to understand, and to facilitate managers to quickly grasp key information.
[0165] It should be understood that, in the embodiments of this application, statistical analysis is used to identify patterns and problems in reimbursement data, providing data support for management decisions. Simultaneously, the visualization report presents complex data in intuitive graphics, improving information transmission efficiency and helping managers make quick decisions. Furthermore, regular statistical analysis allows for monitoring the effectiveness of reimbursement management, timely identification of problems, and adjustment of management strategies, achieving continuous optimization of reimbursement management.
[0166] As an example, a data review module can also be deployed on the server side. The input of the data review module is monthly / quarterly approval data (compliance rate, anomaly type, scenario distribution, and manual intervention rate); the output of the data review module is a data visualization report and suggestions for system optimization.
[0167] The data review module's processing flow can be as follows:
[0168] 1) Key statistical indicators: compliance rate of each scenario, proportion of abnormal types, and distribution of reasons for manual intervention.
[0169] 2) Identify loopholes in the system through trend analysis (e.g., "the compliance rate of remote work subsidies is only 60% due to ambiguity in the rules").
[0170] 3) Generate optimization suggestions (such as "clarify the standard for defining the number of days for remote work subsidies").
[0171] 4) Data is displayed in line charts and pie charts through the data visualization module, and can be exported by administrators.
[0172] It should be understood that reviewing and analyzing expense reimbursement data can promptly identify loopholes and problems in the system, providing a basis for system optimization and improving the company's financial management level. At the same time, data visualization reports allow administrators to intuitively understand the approval process and make quick decisions.
[0173] In one possible implementation, the server in this embodiment may also deploy a knowledge graph module. The inputs to the knowledge graph module are reimbursement rules, historical approval cases, and external business data (project / customer information); the outputs of the knowledge graph module are a structured knowledge graph and case matching results.
[0174] The construction process for a knowledge graph module can be as follows:
[0175] 1) Use a pre-defined graph database to define entities (rules, cases, customers, projects, expense claimants) and relationships ("applicable scenarios for rules", "associated scenarios for cases", "associated projects for customers").
[0176] 2) Synchronize expense reimbursement data from the enterprise system through data acquisition technology, and extract key information (scenario, rules, opinions) from historical approval cases through NLP technology.
[0177] 3) Provide a case matching interface, input scenario information and output similar approval cases (such as matching similar cases in the past 3 months for the "cross-regional customer visit" scenario).
[0178] 4) Supports an intelligent question-and-answer module to respond to approvers' questions such as "accommodation standards for this scenario," thereby achieving efficient transmission of approval knowledge.
[0179] It should be understood that structured knowledge graphs facilitate the storage, retrieval, and management of knowledge, thereby improving the efficiency of knowledge utilization. Simultaneously, the case matching interface and intelligent question-answering module can quickly provide approvers with similar approval cases and rule information, assisting in approval decisions and achieving efficient transfer of approval knowledge.
[0180] Based on the reimbursement approval method provided in the above embodiments, see [link to relevant documentation]. Figure 2 This application also provides a schematic diagram of the structure of a reimbursement approval device.
[0181] Combination Figure 2 As shown, the reimbursement approval device 20 provided in this application embodiment includes:
[0182] The acquisition unit 21 is used to acquire the target reimbursement data to be processed; wherein, the target reimbursement data includes the target reimbursement text, the target scenario tag and the target project information;
[0183] Extraction unit 22 is used to extract keywords from the target reimbursement text to obtain target keywords;
[0184] Determining unit 23 is used to determine the target scene library from the preset scene database based on the target scene label;
[0185] The scenario determination unit 24 is used to determine the target business scenario and target scenario weight corresponding to the target reimbursement text based on the target scenario library, target keywords and target project information;
[0186] Approval rule determination unit 25 is used to determine the target approval rule based on the target business scenario and the target scenario weight;
[0187] Approval unit 26 is used to approve target reimbursement data according to target approval rules and obtain target approval results corresponding to target reimbursement data.
[0188] In one possible implementation, the approval rule determination unit is used for:
[0189] Determine n rules based on the target business scenario; n is an integer greater than or equal to 1.
[0190] Sort the n rules according to the target scenario weight;
[0191] The target approval rule is determined based on the sorted n rules.
[0192] In one possible implementation, if the target approval result is characterized as abnormal, it further includes: an abnormality report generation unit, used for:
[0193] Retrieve historical expense reimbursement data associated with the target expense reimbursement data;
[0194] Identify abnormal characteristics of the target based on target reimbursement data and historical reimbursement data;
[0195] Determine the alarm level of the target expense reimbursement data based on the target's abnormal characteristics;
[0196] Determine the anomaly tracing map corresponding to the target expense reimbursement data based on the target expense reimbursement data;
[0197] A target anomaly report is generated based on the alarm level and the anomaly source map.
[0198] In one possible implementation, the apparatus further includes: a visualization report generation unit, used for:
[0199] Retrieve k expense reimbursement data within a preset time period; where k is an integer greater than or equal to 1;
[0200] Calculate the compliance rate and the percentage of abnormal types based on k expense reimbursement data;
[0201] Generate a target statistics table based on compliance rate and the proportion of anomaly types;
[0202] Analyze the target statistics table to obtain the target analysis results;
[0203] The target visualization report is determined based on the target statistics table and target analysis results.
[0204] It should be noted that the reimbursement approval device provided in this application embodiment has the same beneficial effects as the reimbursement approval method provided in the above embodiments, so it will not be described again.
[0205] In one possible implementation, see Figure 3 The figure is a schematic diagram of a control device provided in an embodiment of this application.
[0206] The control device may include a memory 311 and a processor 312. For example... Figure 3As shown, the memory can be random access memory (RAM), flash memory, read-only memory (ROM), EPROM, non-volatile read-only memory (Electronic Programmable ROM), registers, hard disks, removable disks, etc.
[0207] The memory 311 can store computer instructions. When the computer instructions stored in the memory 311 are executed by the processor 312, the processor 312 can be used to execute data processing methods. The memory 311 can also store data, such as target approval rules and other information involved in the above embodiments.
[0208] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape) or a semiconductor medium (e.g., solid-state disk (SSD)).
[0209] This application also provides a readable storage medium for storing the methods provided in the above embodiments. Examples include random access memory (RAM), flash memory, read-only memory (ROM), EPROM, non-volatile read-only memory (EPROM), registers, hard disks, removable disks, or any other form of storage medium in the art.
[0210] In the embodiments of this application, the terms "first" and "second" (if they exist) are used only as name identifiers and do not represent the order of first and second.
[0211] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. Regarding the methods disclosed in the embodiments, since they correspond to the product embodiments disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the description of the product embodiments.
[0212] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A reimbursement approval method, characterized in that, The method includes: Obtain the target expense reimbursement data to be processed; wherein, the target expense reimbursement data includes target expense reimbursement text, target scenario tags, and target project information; The target reimbursement text is processed to extract keywords, thereby obtaining target keywords; The target scene library is determined from the preset scene database based on the target scene tags; Based on the target scenario library, the target business scenario and target scenario weight corresponding to the target reimbursement text are determined according to the target keywords and the target project information; The target approval rules are determined based on the target business scenario and the target scenario weight. The target reimbursement data is approved according to the target approval rules to obtain the target approval result corresponding to the target reimbursement data.
2. The reimbursement approval method according to claim 1, characterized in that, The step of determining the target approval rule based on the target business scenario and the target scenario weight includes: Based on the target business scenario, n rules are determined; where n is an integer greater than or equal to 1. Sort the n rules according to the target scene weight; The target approval rule is determined based on the sorted n rules.
3. The reimbursement approval method according to claim 1, characterized in that, If the target approval result is characterized as abnormal, it also includes: Obtain the historical reimbursement data associated with the target reimbursement data; Based on the target reimbursement data and the historical reimbursement data, determine the target anomaly characteristics; Determine the alarm level of the target expense reimbursement data based on the target's abnormal characteristics; Based on the target expense reimbursement data, determine the anomaly tracing map corresponding to the target expense reimbursement data; A target anomaly report is generated based on the alarm level and the anomaly source map.
4. The reimbursement approval method according to claim 1, characterized in that, The method further includes: Obtain k expense reimbursement data within a preset time period; wherein, k is an integer greater than or equal to 1; Calculate the compliance rate and the percentage of abnormal types based on the k reimbursement data; A target statistics table is generated based on the compliance rate and the percentage of the anomaly types. The target statistics table is analyzed to obtain the target analysis results; A target visualization report is determined based on the target statistics table and the target analysis results.
5. A reimbursement approval device, characterized in that, The device includes: The acquisition unit is used to acquire the target reimbursement data to be processed; wherein, the target reimbursement data includes target reimbursement text, target scenario tags, and target project information; The extraction unit is used to perform keyword extraction processing on the target reimbursement text to obtain target keywords; The determining unit is used to determine a target scene library from a preset scene database based on the target scene label; The scenario determination unit is used to determine the target business scenario and target scenario weight corresponding to the target reimbursement text based on the target scenario library, the target keywords, and the target project information; An approval rule determination unit is used to determine a target approval rule based on the target business scenario and the target scenario weight. An approval unit is used to approve the target reimbursement data according to the target approval rules and obtain the target approval result corresponding to the target reimbursement data.
6. The reimbursement approval device according to claim 5, characterized in that, The approval rule determination unit is used for: Based on the target business scenario, n rules are determined; where n is an integer greater than or equal to 1. Sort the n rules according to the target scene weight; The target approval rule is determined based on the sorted n rules.
7. The reimbursement approval device according to claim 5, characterized in that, If the target approval result is characterized as abnormal, the system further includes: an abnormality report generation unit, used for: Obtain the historical reimbursement data associated with the target reimbursement data; Based on the target reimbursement data and the historical reimbursement data, determine the target anomaly characteristics; Determine the alarm level of the target expense reimbursement data based on the target's abnormal characteristics; Based on the target expense reimbursement data, determine the anomaly tracing map corresponding to the target expense reimbursement data; A target anomaly report is generated based on the alarm level and the anomaly source map.
8. The reimbursement approval device according to claim 5, characterized in that, The device further includes: a visualization report generation unit, used for: Obtain k expense reimbursement data within a preset time period; wherein, k is an integer greater than or equal to 1; Calculate the compliance rate and the percentage of abnormal types based on the k reimbursement data; A target statistics table is generated based on the compliance rate and the percentage of the anomaly types. The target statistics table is analyzed to obtain the target analysis results; A target visualization report is determined based on the target statistics table and the target analysis results.
9. A control device, characterized in that, It includes a processor and a memory, the memory being used to store programs, instructions, or code, and the processor being used to execute the programs, instructions, or code in the memory to complete the reimbursement approval method as described in any one of claims 1-4.
10. A computer-readable storage medium, characterized in that, The device contains a computer program that is loaded by a processor to execute the reimbursement approval method as described in any one of claims 1-4.