Intelligent processing method for automatic settlement of review labor reward of review expert

Abstract

The invention discloses an intelligent processing method for automatic settlement of review labor reward of a review expert, and belongs to the technical field of business data processing, and the method comprises the steps: obtaining text revision trajectory data and cooperative interaction data of the review expert, and generating a review behavior feature vector; performing recessive workload analysis on the review behavior feature vector to generate a quality weight factor; obtaining project environment parameters, and calculating to obtain a reward amount according to the project environment parameters, the quality weight factor and a preset settlement coefficient; performing exception verification on the reward amount, if the verification is passed, executing payment operation, and otherwise, triggering a manual rechecking process; and updating the settlement coefficient according to the feedback data after payment. According to the method, the reward is dynamically calculated by quantitatively reviewing the behavior characteristics and fusing the environmental parameters, and an exception verification and closed-loop optimization mechanism is established, so that fair and efficient full-process automatic settlement can be realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of business data processing, in particular to an intelligent processing method for automatic settlement of review expert review labor remuneration. BACKGROUND

[0002] In the professional fields of scientific research projects and fund review, review experts participate in the review work by revising document content and making suggestions, and the settlement of their labor remuneration needs to consider the review quality, workload and project complexity. The traditional settlement process relies on manual statistics and experience judgment, which is low in efficiency and difficult to standardize.

[0003] The prior art usually calculates the remuneration by using a fixed formula, for example, paying an amount according to the number of review documents or a preset standard, and some systems introduce a simple time factor or a basic quality score as an adjustment parameter. In cross-border payment scenarios, the exchange rate needs to be manually converted, and the handling of disputed projects relies on manual identification and negotiation.

[0004] However, manual accounting is inefficient and prone to errors, and the fixed formula cannot accurately quantify the implicit value of the review behavior; there is a lack of dynamic compensation mechanism for cross-border exchange rate fluctuations, urgent task time limits, non-working hours, etc.; the remuneration calculation result lacks effective verification means, making it difficult to identify data anomalies or model bias; the settlement parameters are fixed for a long time and cannot be self-adaptively optimized according to expert feedback or behavior changes. SUMMARY

[0005] To solve the above problems, the present application provides an intelligent processing method for automatic settlement of review expert review labor remuneration, which dynamically calculates the remuneration by quantifying the review behavior characteristics and fusing environmental parameters, and establishes an abnormality verification and closed-loop optimization mechanism, thereby achieving fair and efficient full-process automatic settlement.

[0006] The above-mentioned object can be achieved by the following scheme: An intelligent processing method for automatic settlement of review expert review labor remuneration, comprising: obtaining text revision track data and collaborative interaction data of a review expert, and generating a review behavior feature vector; performing implicit workload analysis on the review behavior feature vector to generate a quality weight factor; obtaining project environmental parameters, and calculating a remuneration amount according to the project environmental parameters, the quality weight factor and a preset settlement coefficient; performing abnormality verification on the remuneration amount, and if the verification is passed, performing a payment operation, otherwise triggering a manual review process; updating the settlement coefficient according to feedback data after payment.

[0007] Optionally, the generating the review behavior feature vector comprises: collecting text revision track data of the review expert on the review document, and extracting the number of modification points per thousand Chinese characters as a revision density feature; capturing the collaborative interaction data of the review expert, and counting the number of times that the review suggestion is cited by other experts as a collaborative heat value; and combining the revision density feature and the collaborative heat value to generate the review behavior feature vector.

[0008] Optionally, the generating the quality weight factor comprises: calculating a basic quality score according to the revision density feature; calculating a collaborative contribution score according to the collaborative heat value; and calculating the quality weight factor by using the basic quality score, the collaborative contribution score, and a preset weight ratio.

[0009] Optionally, the calculating the remuneration amount comprises: obtaining a project environment parameter, and filtering, according to the project environment parameter, a review item whose number of opinion-divergent experts is greater than a preset number threshold to construct a controversial project set; calculating a controversial additional correction coefficient according to the controversial project set; when a cross-border payment identifier is detected, calling a real-time exchange rate fluctuation buffer pool to perform exchange rate compensation calculation to obtain an exchange rate compensation value; and calculating the remuneration amount by using the controversial additional correction coefficient, the exchange rate compensation value, the quality weight factor, and a preset settlement coefficient.

[0010] Optionally, the method further comprises: obtaining a project deadline and a current time according to the project environment parameter, and calculating a difference between the project deadline and the current time to obtain a project deadline time value; calculating a project urgency coefficient by using the project deadline time value; obtaining a review operation timestamp, and generating a cross-time-zone review identifier when the review operation timestamp is in a preset night time period; and modifying the quality weight factor by using the project urgency coefficient when the cross-time-zone review identifier is identified.

[0011] Optionally, the performing abnormality verification on the remuneration amount comprises: obtaining historical review behavior data and corresponding historical remuneration of a review expert, wherein the historical review behavior data comprises historical review start time, historical end time, and historical review document quantity; calculating a historical review efficiency by using the historical review document quantity; constructing a relationship mapping table of the historical review efficiency and the corresponding historical remuneration; obtaining a total number of review documents completed within a preset time window from a current time point of the review expert, and calculating a current review efficiency; obtaining a remuneration reference value according to the current review efficiency and the relationship mapping table; calculating a remuneration difference value between the remuneration amount and the remuneration reference value; and generating an abnormality verification identifier when the remuneration difference value is greater than a preset remuneration threshold.

[0012] Optionally, the triggering the manual review process comprises: triggering the manual review process when the abnormality check flag exists; performing difference analysis on the reward reference value and the reward amount; and adjusting the weight proportion according to the difference analysis result.

[0013] Optionally, the updating the settlement coefficient according to the post-payment feedback data comprises: obtaining satisfaction feedback data of the review expert on the reward amount, obtaining a satisfaction score and storing the satisfaction score in a satisfaction database; judging whether the settlement coefficient needs to be updated based on the satisfaction database; when the settlement coefficient needs to be updated, calculating a satisfaction coefficient by using data in the satisfaction database; calculating a settlement correction factor according to the satisfaction coefficient and the quality weight factor; and updating the settlement coefficient by using the settlement correction factor.

[0014] Optionally, the method further comprises: after the payment operation is completed, continuously collecting expert behavior change data in a preset period, the expert behavior change data including response speed of a subsequent review project, review document acceptance rate and collaboration interaction frequency; quantifying the expert behavior change data and calculating a comprehensive participation index; and when the comprehensive participation index is lower than a preset baseline, updating the collection dimension of the review behavior feature vector.

[0015] Based on the same inventive concept, the application further provides an intelligent processing system for automatic settlement of review expert review labor reward, which comprises: a review behavior collection module, configured to obtain text revision track data and collaborative interaction data of a review expert, and generate a review behavior feature vector; an intelligent value analysis module, configured to perform implicit workload analysis on the review behavior feature vector, and generate a quality weight factor; a reward calculation module, configured to obtain project environment parameters, and calculate a reward amount according to the project environment parameters, the quality weight factor and a preset settlement coefficient; a self-check payment module, configured to perform abnormality check on the reward amount, and if the check is passed, perform a payment operation, otherwise trigger a manual review process; and a feedback optimization module, configured to update the settlement coefficient according to post-payment feedback data.

[0016] Compared with the prior art, the application has the following advantages: 1. The application generates a feature vector by quantifying the text revision track and collaborative interaction behavior of a review expert, realizes objective evaluation of implicit workload, solves the problem that review quality is difficult to accurately measure in traditional manual accounting, and significantly improves the scientificity and fairness of reward calculation. 2. The application dynamically adjusts a reward calculation model in combination with project environment parameters, automatically adapts a correction coefficient for complex scenes such as controversial projects, cross-border payments and urgent tasks, enhances the adaptability of the system to multi-dimensional influence factors, and ensures the rationality of reward distribution in different scenes. 3. This invention establishes an anomaly verification mechanism based on historical behavior data. By comparing the remuneration amount with the efficiency model benchmark in real time, it identifies deviations and triggers manual review, effectively reducing the risk of calculation errors and ensuring the accuracy and credibility of automated settlement results. 4. This invention uses post-payment feedback data and behavior change data to continuously optimize the settlement coefficient and feature collection dimensions, forming a closed-loop learning mechanism, so that the compensation model evolves dynamically with expert satisfaction and participation, thereby improving the long-term stability and incentive effect of the system.

[0017] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 The present invention is a flowchart of an intelligent processing method for automated settlement of evaluation expert evaluation service remuneration according to an embodiment of the present invention.

[0020] Figure 2 Schematic diagram of the composition of the remuneration amount according to an embodiment of the present invention.

[0021] Figure 3 The present invention is a schematic diagram of the structure of an intelligent processing system for automatic settlement of evaluation expert evaluation service remuneration according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings 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 making creative efforts shall fall within the scope of protection of the present invention.

[0023] Reference Figure 1An embodiment of the present application provides an intelligent processing method for reviewing expert review labor remuneration automatic settlement, which can realize fair and efficient full-process automatic settlement by quantifying review behavior characteristics, dynamically calculating remuneration by fusing environmental parameters, and establishing an abnormality checking and closed-loop optimization mechanism.

[0024] The method of the embodiment specifically comprises: acquiring text revision track data and collaborative interaction data of the review expert, and generating a review behavior characteristic vector; performing implicit workload analysis on the review behavior characteristic vector, and generating a quality weight factor; acquiring project environmental parameters, and calculating a remuneration amount according to the project environmental parameters, the quality weight factor and a preset settlement coefficient; performing abnormality checking on the remuneration amount, and if the checking is passed, performing a payment operation, otherwise triggering a manual review process; updating the settlement coefficient according to feedback data after payment.

[0025] Specifically, the remuneration amount is as shown in Figure 2 achieved by multi-dimensional data fusion and dynamic modeling. First, the text revision track and collaborative interaction data of the review expert are collected and quantified into a calculable review behavior characteristic vector, which objectively represents the implicit workload. Then, intelligent analysis is performed on the characteristic vector to extract quantified indicators of review quality and collaboration value, and a quality weight factor is generated. A remuneration calculation model is constructed by combining project environmental parameters and a preset settlement coefficient, realizing multi-factor dynamic compensation. Through abnormality checking based on historical behavior data, a dual protection mechanism of automatic payment and manual review is formed. Finally, the settlement coefficient is updated in a closed loop based on feedback data after payment, so that the model is continuously optimized. The intelligent level of review remuneration settlement is significantly improved. The implicit review value that is difficult to evaluate in the traditional way is accurately quantified through the behavior characteristic vector, and the subjective bias of manual calculation is eliminated. The dynamic remuneration model automatically adapts to complex scenarios such as controversial projects and cross-border payments, ensuring fair compensation for multi-dimensional intellectual contributions. The abnormality checking mechanism effectively prevents calculation errors by modeling based on historical data, ensuring the reliability of the settlement results. The closed-loop optimization function enables the system to evolve continuously with expert feedback, maintaining the fairness of settlement in the long term, significantly reducing management costs and improving review efficiency.

[0026] Optionally, the generation of the review behavior characteristic vector comprises: collecting text revision track data of the review expert on the review document, and extracting the number of modification points per thousand characters as a revision density feature; capturing collaborative interaction data of the review expert, and counting the number of times that the review suggestions are cited by other experts as a collaborative heat value; combining the revision density feature and the collaborative heat value to generate the review behavior characteristic vector.

[0027] Specifically, first, the text revision track data of the review document by the review expert is collected through the document version control system, and the text revision track data includes document modification position markers and modification contents. The total number of the document modification position markers is extracted as the modification point number, and the total number of words of the review document is obtained. The modification point number per thousand words is calculated as the revision density feature, and for the revision density feature value , , In the formula, is the modification point number, is the total number of words of the document. The modification point number is integer type data, and the total number of words of the document is automatically obtained by a text analysis tool. Second, the collaborative interaction data is captured through the review system log, and the collaborative interaction data includes the reference records of the review suggestions by other experts. The cumulative number of times that each review suggestion of the review expert is cited by other experts in the review process is counted as the collaborative heat value , and the collaborative heat value is a non-negative integer. Finally, the numerical revision density feature and the collaborative heat value are combined in a fixed order to form a two-dimensional vector to generate a review behavior feature vector, which is in the form of .

[0028] Exemplarily, a review expert participates in the review of a project, and the system collects the total number of words of the review document as 5000 words. It is detected that the expert marks 15 modification points in the document. The revision density feature is calculated as (15 / 5000) * 1000 = 3. At the same time, the system captures that 3 review suggestions proposed by the expert are cited by other experts, of which the first suggestion is cited 2 times, the second suggestion is cited 1 time, and the third suggestion is not cited. The cumulative number of citations is 3 times, so the collaborative heat value is 3. The revision density feature value 3 and the collaborative heat value 3 are combined to generate a review behavior feature vector [3, 3]. The beneficial effect of this verification example is that the text revision density and the collaborative contribution degree are quantified objectively, the implicit review workload which is difficult to measure in the traditional way is converted into a calculable feature vector, which provides a standardized input for subsequent reward calculation, and at the same time promotes the review expert to pay attention to the suggestion quality and the collaboration value in the review process, and improves the overall efficiency of the review system.

[0029] Optionally, the generating the quality weight factor comprises: calculating a basic quality score according to the revision density feature; calculating a collaborative contribution score according to the collaborative heat value; calculating the quality weight factor by using the basic quality score, the collaborative contribution score and a preset weight ratio.

[0030] Specifically, first, the revision density feature and the synergy heat value in the review behavior feature vector are obtained; the linear conversion method is used to map the revision density feature value to the preset score interval, specifically, the revision density feature is multiplied by the preset first conversion coefficient to obtain the basic quality score : , wherein is the preset first conversion coefficient, a constant parameter determined by the system administrator according to historical review data statistical analysis, used to map the typical revision density value to the 0-100 score interval; the linear conversion method is used to map the synergy heat value to the same score interval at the same time, specifically, the synergy heat value is multiplied by the preset second conversion coefficient to obtain the synergy contribution score : , wherein is the preset second conversion coefficient, which is also a constant parameter calibrated by historical data; then a preset weight ratio parameter is called, which includes the basic quality weight and the synergy contribution weight; finally, the quality weight factor is obtained by weighted calculation : , wherein, is the basic quality weight, is the synergy contribution weight.

[0031] Exemplarily, the revision density feature is 4 and the synergy heat value is 5 in a certain review; the system preset first conversion coefficient is 20, then the basic quality score is 80; the preset second conversion coefficient is 10, then the synergy contribution score is 50; the preset weight ratio , ; the quality weight factor is 68. The heterogeneous behavior data is unified into a comparable score system through the standardized conversion coefficient, and the multi-dimensional quantitative fusion of the review quality is realized by using the configurable weight ratio, which objectively reflects the individual text revision depth of the expert and embodies the team collaboration value, providing a scientific basis for the reward calculation. At the same time, the flexibility of the weight parameter can adapt to the differentiated evaluation needs of the review behavior of different project types.

[0032] Optionally, the calculated reward amount comprises: Obtain the project environment parameters, and filter the review items whose number of opinion difference experts is greater than a preset number threshold to construct a controversial project set; Calculate a controversial additional correction coefficient according to the controversial project set; When the cross-border payment identifier is detected, call the real-time exchange rate fluctuation buffer pool to perform exchange rate compensation calculation to obtain an exchange rate compensation value; The compensation amount is calculated by using the dispute additional correction coefficient, the exchange rate compensation value, the quality weight factor and a preset settlement coefficient.

[0033] Specifically, first, the project environment parameters are acquired, including the total number of experts in the review project, the number of opinion-divergent experts and the project type identifier; a preset number threshold is set, and when it is detected that the number of opinion-divergent experts corresponding to a review item is greater than the preset number threshold, the review item is included in the dispute project set; the dispute additional correction coefficient is calculated : , wherein is the actual number of divergent experts, is the preset number threshold, is the dispute intensity coefficient and , is determined by the historical dispute project compensation adjustment ratio mean; the country code field in the payment request is detected synchronously, and when the country code is inconsistent with the expert registration code, the cross-border payment identifier is activated; if the cross-border payment identifier exists, the real-time exchange rate fluctuation buffer pool is accessed, the buffer pool stores the recent hourly exchange rate standard deviation , the exchange rate compensation value is calculated: , wherein is the buffer factor (default value is 0.05), is the benchmark compensation (i.e. the compensation intermediate value without calculating the dispute correction), is the recent hourly exchange rate standard deviation, which is obtained by an exchange market API; finally, for the compensation amount , there is: , wherein, is a preset settlement coefficient, and the dimensions of various parameters are unified as currency units.

[0034] Exemplarily, a certain cross-border review project has a preset quantity threshold of 3, detects that there are 5 expert opinion differences in a certain review item, takes the controversy intensity coefficient as 0.15, calculates the controversy additional correction coefficient as 1.25; the project activates the cross-border payment identifier, queries the exchange rate standard deviation of the exchange rate fluctuation buffer pool as 0.8 (T = 24 hours), takes the preset benchmark remuneration as 2000 yuan, the buffer factor as 0.05, and calculates the exchange rate compensation value as 80 yuan; assuming that the quality weight factor is 1.2 and the settlement coefficient is 1.0, the final remuneration amount is 3080 yuan. By dynamically constructing a controversial item set to identify high-value review work, using a step-by-step correction coefficient to quantify the controversy processing difficulty, and using an exchange rate compensation mechanism to eliminate cross-border payment fluctuation risks, the remuneration calculation reflects the difference in intellectual contribution in a complex project environment and guarantees payment stability, significantly improving the fairness of settlement in cross-border multi-expert collaboration scenarios.

[0035] Optionally, the method further comprises: According to the project environment parameters, the project deadline and the current time are obtained, and the difference between the project deadline and the current time is calculated to obtain a project deadline time value; Using the project deadline time value, a project urgency coefficient is calculated; An inter-time-zone review identifier is generated when the review operation timestamp is in a preset night time period; When the cross-time-zone review identifier is identified, the quality weight factor is corrected by the project urgency coefficient.

[0036] Specifically, first, the project deadline and the current time are extracted from the project environment parameters, the project environment parameters are stored in the project database, the project deadline is the scheduled completion time point of the project, and the current time is the time obtained by the system real-time clock; the difference between the project deadline and the current time is calculated to obtain a project deadline time value; then the project deadline time value is used to calculate a project urgency coefficient : , wherein is a preset emergency sensitivity coefficient, with units of hours, and a typical value is determined by analyzing the correlation between time pressure and review efficiency in historical projects, is a preset smoothing constant, with units of hours, used to avoid division by zero errors, and is set by the smallest time unit, is a time value of the project deadline; a review operation timestamp is obtained at the same time, and the review operation timestamp is a specific time point of the expert submitting the review operation recorded by the review system log; a preset night time period is defined as 22:00 to 06:00 of the local time and is stored in a system configuration file; when the review operation timestamp falls within the night time period, a cross-time-zone review identifier is generated, and the identifier is a Boolean flag; if the cross-time-zone review identifier is identified, the quality weight factor is modified, and for the modified quality weight factor , there are: .

[0037] Exemplarily, the project deadline time value is 24 hours, the preset emergency sensitivity coefficient is 24 hours, the smoothing constant is 1 hour, the project emergency degree coefficient is 1.96, the review operation timestamp is 2023-10-04-23:30, which is in the preset night time period, the cross-time-zone review identifier is generated, the original quality weight factor is 1.0, and the modified quality weight factor is 1.96. By dynamically quantifying the project emergency degree and identifying the non-working time review behavior, the compensation calculation weight is automatically adjusted, the extra effort of the expert in the high-pressure or cross-time-zone scenario is effectively compensated, the adaptability of the review system to the complex working environment is enhanced, the enthusiasm of the expert and the completion rate of the review task are improved, and at the same time, the compensation distribution is more fair and reasonable.

[0038] Optionally, the abnormality verification on the compensation amount comprises: obtaining historical review behavior data and corresponding historical compensation of a review expert, wherein the historical review behavior data comprises historical review start time, historical end time and historical number of review documents; calculating a historical review efficiency by using the historical number of review documents; constructing a relationship mapping table of historical review efficiency and corresponding historical compensation obtaining a total number of review documents completed within a preset time window forward from a current time point of the review expert, and calculating a current review efficiency; obtaining a compensation reference value according to the current review efficiency and the relationship mapping table; calculating a compensation difference value between the compensation amount and the compensation reference value; generating an abnormality verification flag when the compensation difference value is greater than a preset compensation threshold.

[0039] Specifically, first, historical review behavior data set of a review expert is obtained from a review system database, the data set includes historical review start time, historical end time and historical number of review documents; a historical review efficiency value is calculated : , in The number of historical review documents (integer type), The historical review duration (in hours) is obtained by subtracting the historical start time from the historical end time and converting it into hours; a relationship mapping table is constructed based on all historical review behavior data. As the horizontal axis, the corresponding historical compensation is the vertical axis, and the linear regression method is used to establish the functional relationship. ,have: ; Where, For the current review efficiency, synchronously obtain the preset time window forward from the current time point The total number of review documents completed by the expert within a period (e.g., 30 days) , calculate the current review efficiency : , in For a fixed time span (unit: day), it needs to be converted to hourly units to keep the dimension consistent; Substitute the relationship mapping table function to obtain the reward benchmark value ; Calculate the difference between the remuneration amount and the remuneration benchmark value, which is an absolute value; when the remuneration difference is greater than the preset remuneration threshold, generate an abnormality check flag, where the remuneration threshold is set according to the standard deviation of historical remuneration fluctuations.

[0040] For example, the historical data of an expert shows that when the review efficiency is 0.5 articles / hour, the corresponding remuneration is 2,000 yuan, and when it is 0.8 articles / hour, the corresponding remuneration is 3,000 yuan. ; Current detection 576 reviews were completed within 720 days. Articles / hour, calculated remuneration base value is 2,750 yuan; actual remuneration amount is 3,500 yuan, remuneration threshold is 500 yuan, and remuneration difference is 750, which is greater than 500, triggering an abnormal check flag. Through historical behavior modeling to establish a personalized remuneration benchmark system, dynamic detection of significant deviations between current remuneration and efficiency models, effectively identifying remuneration distortions caused by system calculation errors or abnormal behavioral data, providing accurate basis for manual review, ensuring the accuracy and anti-interference ability of the remuneration settlement system, and maintaining experts' trust in the automated settlement system.

[0041] Optionally, the triggering of the manual review process includes: When the abnormal check mark exists, a manual review process is triggered; differing the reward amount from the reward benchmark value; adjusting the weight proportion according to the difference analysis result.

[0042] Specifically, when the system detects an abnormality check flag, an artificial review process is automatically triggered. The abnormality check flag is generated when the reward difference between the reward amount and the reward benchmark value exceeds a preset reward threshold. The reward benchmark value is obtained by querying a historical review behavior data set, which includes historical review start time, historical review end time, and historical review document quantity; the historical review efficiency is calculated as the historical review document quantity divided by the historical review duration, wherein the historical review duration is obtained by subtracting the historical start time from the historical end time and converting it into hours; based on a relationship mapping table between the historical review efficiency and the corresponding historical reward, a reward benchmark function is established using a linear regression method; the total number of review documents completed within a preset time window from the current time point is obtained, and the current review efficiency is calculated as the total number of review documents divided by the number of hours in the time window; the current review efficiency is substituted into the reward benchmark function to output the reward benchmark value. After triggering the artificial review process, the system performs difference analysis on the reward benchmark value and the reward amount, including calculating the absolute value of the reward difference and identifying the source of the difference, such as abnormality in review behavior feature vector collection or parameter configuration deviation. The difference analysis result outputs a difference report containing the deviation type and adjustment suggestions. According to the difference analysis result, the weight proportion is automatically adjusted, including the basic quality weight and the synergistic contribution weight, which are used for subsequent quality weight factor calculation. The adjustment of the weight proportion is based on the deviation type, such as reducing the basic quality weight for revision density feature abnormality or reducing the synergistic contribution weight for synergistic heat value abnormality, and the adjustment amplitude is determined through a preset rule table that stores the mapping relationship between the deviation type and the weight adjustment coefficient. The automatic settlement is combined with manual intervention through the abnormality triggering mechanism, the difference analysis locates the root cause of the system deviation, and the weight proportion is dynamically optimized to improve the model adaptability; the technical effect is to enhance the fault tolerance of the reward settlement system, reduce payment errors caused by data abnormalities or parameter inaccuracies, improve review fairness and expert satisfaction, and continuously optimize system performance through parameter self-adaptive adjustment, ensuring long-term operation stability.

[0043] Exemplarily, in a certain review project, the system detects that the reward amount is 3500 yuan, the reward benchmark value is 2750 yuan, the reward difference absolute value is 750 yuan, and the preset reward threshold value is 500 yuan. The reward difference exceeds the threshold value to trigger the abnormal verification flag. After the manual review process is started, the reviewer analyzes the difference report and finds that the collaborative heat value collection module has a log analysis error, which leads to the virtual high value of the collaborative heat value, and further causes the abnormal increase of the reward amount. According to the difference analysis result, the system automatically reduces the collaborative contribution weight from 0.4 to 0.35. Through the accurate identification of data collection defects by manual review, the calculation deviation of the reward is corrected in time to avoid payment errors and maintain the rights and interests of experts; at the same time, the weight proportion is dynamically adjusted to optimize the subsequent calculation model, to prevent similar problems from occurring, to improve the reliability and trustworthiness of the automatic settlement system, and to strengthen the quality orientation of the review.

[0044] Optionally, the updating the settlement coefficient according to the feedback data after payment comprises: obtaining satisfaction feedback data of the review experts on the reward amount, obtaining a satisfaction score and storing it in a satisfaction database; judging whether it is necessary to update based on the satisfaction database; when it is necessary to update, calculating a satisfaction coefficient by using the data in the satisfaction database; calculating a settlement correction factor according to the satisfaction coefficient and the quality weight factor; updating the settlement coefficient by using the settlement correction factor.

[0045] Specifically, after the payment operation is completed, a settlement coefficient updating process is executed. First, satisfaction feedback data of the review experts on the reward amount is obtained through a feedback interface of the review system. The data contains a numerical score submitted by the expert, and the score range is 1 to 10 points, indicating the satisfaction level. The score is extracted and stored in a satisfaction database, which is a structured storage system recording expert ID, score value and time stamp. Second, whether it is necessary to update the settlement coefficient is judged based on the satisfaction database. Specifically, the increment proportion of the new data in the database is calculated. If the number of new score records exceeds a preset threshold value or the average score change absolute value in the recent preset period is greater than a preset change threshold value, it is determined that the settlement coefficient needs to be updated. When it is necessary to update, a satisfaction coefficient is calculated by using the data in the satisfaction database. The satisfaction coefficient is equal to the arithmetic mean of all score values in the database, which is obtained by querying the database and calculating the sum divided by the number of records. Then, a settlement correction factor is calculated according to the satisfaction coefficient and the quality weight factor is a preset target satisfaction value, which is set as a fixed constant by the system administrator according to historical satisfaction data statistical analysis, ​​​is the satisfaction coefficient. Finally, the settlement coefficient is updated by using the settlement correction factor : , The reward calculation parameters are optimized through expert feedback closed loop, the settlement coefficient is dynamically adjusted by using the satisfaction data, the reward distribution is more in line with the expectations of experts, the adaptability and fairness of the reward system are improved, the enthusiasm of experts is enhanced, the disputes caused by settlement deviation are reduced, and the long-term stable operation of the system is promoted.

[0046] Exemplarily, after a payment is completed, the system obtains the satisfaction scores of three experts for the amount of reward as 7, 8 and 9 respectively, and stores the scores in the satisfaction database. A preset new data increment threshold is 2, and 3 new records exceed the threshold, triggering the update judgment. The satisfaction coefficient is calculated as (7+8+9) / 3=8. A preset target satisfaction is 8.5, a quality weight factor is 1.0, a settlement correction factor C is calculated as 8 / 8.5≈0.94. The current settlement coefficient is 1.0, and the updated settlement coefficient is 0.94. Through automatic collection and analysis of expert feedback data, the satisfaction change is accurately identified and the settlement parameters are timely optimized, so that the reward calculation model continuously adapts to the needs of experts, the trust and incentive effect of the review system are effectively improved, the problem of unfair reward caused by static parameters is avoided, the manual maintenance cost is reduced, and the self-optimization ability of the system is strengthened.

[0047] Optionally, the method further comprises: After the payment operation is completed, the expert behavior change data in a preset period is continuously collected, and the expert behavior change data includes the response speed of the subsequent review project, the review document acceptance rate and the cooperation interaction frequency; The expert behavior change data is quantified, and a comprehensive participation index is calculated; When the comprehensive participation index is lower than a preset baseline, the collection dimension of the review behavior feature vector is updated.

[0048] Specifically, after the payment operation is completed, the system starts the expert behavior change monitoring process, and continuously collects the expert behavior change data in the preset period. The preset period is configured by the system administrator, and the default value is 30 days. The expert behavior change data includes response speed, review document acceptance rate and collaboration interaction frequency. The response speed is defined as the time difference from receiving the review invitation to starting the review operation, in hours, and the average value is obtained by calculating the time stamp in the review system log. The review document acceptance rate is defined as the number of review documents accepted by the expert divided by the total number of review invitations sent by the system. This value is a dimensionless proportion, and is obtained by statistical database of review invitation records. The collaboration interaction frequency is defined as the number of interactions between the expert and other experts in the review process, including comment, reference or reply operations, in times, and the cumulative value is obtained by statistical review collaboration log. The above data is quantitatively processed: the response speed is converted into an efficiency value using the inverse form, the review document acceptance rate is directly used as the quantitative value, and the collaboration interaction frequency is mapped to the preset range using a linear conversion coefficient. The comprehensive participation index is calculated by a weighted formula. For the comprehensive participation index , there is: , wherein, is the response speed quantitative value, the calculation formula is , is the response speed average value, is the preset response conversion coefficient (calibrated by historical data); is the review document acceptance rate, which is dimensionless; is the collaboration interaction frequency quantitative value, the calculation formula is , is the collaboration interaction frequency, is the preset interaction conversion coefficient (calibrated by historical data) , , respectively, the response speed weight, the acceptance rate weight and the interaction frequency weight, are all preset dimensionless weight parameters, and the sum is 1. After the comprehensive participation index is calculated, it is compared with the preset baseline, which is a constant threshold set by the system administrator. When the comprehensive participation index is less than the preset baseline, the collection dimension of the review behavior feature vector is updated, specifically, on the basis of the original revision density feature and the synergy heat value, the response speed or the review document acceptance rate is added as a new dimension, and the updated feature vector form is [revision density feature, synergy heat value, new feature]. The selection of the new feature is based on the dimension corresponding to the lowest quantitative value in the expert behavior change data. In the operation, the data collection module is automatically called to reconfigure the collection logic to ensure that the updated feature vector is used in subsequent reward calculation.

[0049] Exemplarily, after the payment operation is completed, the system collects the behavior change data of a certain review expert in a preset period of 30 days: the average response speed is 48 hours, the review document acceptance rate is 0.85, and the collaboration interaction frequency is 8 times. The preset response conversion coefficient is 0.02, the response speed quantitative value is calculated as 0.02 / 48≈0.0004; the preset interaction conversion coefficient is 0.1, the interaction frequency quantitative value is calculated as 0.1x8=0.8; the review document acceptance rate is directly used as 0.85. The preset weight is the response speed weight 0.3, the acceptance rate weight 0.4, and the interaction frequency weight 0.3, and the comprehensive participation index is calculated as 0.5801. The preset baseline is 0.7, and the comprehensive participation index 0.5801 is less than 0.7, triggering the update operation. Since the response speed quantitative value is the lowest, the response speed is added as a new dimension to the review behavior feature vector, and the updated vector form is [revision density feature, synergy heat value, response speed]. By quantifying the behavior change of the expert after payment and dynamically adjusting the data collection dimension, the system can more comprehensively capture the change of the expert's participation, timely optimize the reward calculation input, avoid the contribution evaluation deviation caused by static characteristics, thereby improving the adaptability and fairness of the review system, enhancing the enthusiasm of the experts for continuous participation, reducing the reward settlement controversy, and promoting the long-term stable improvement of the review quality.

[0050] Based on the same inventive concept, as shown in Figure 3 The present application also provides an intelligent processing system for automatic settlement of review expert review labor remuneration, which comprises: A review behavior acquisition module is used to acquire text revision track data and collaborative interaction data of a review expert, and generate a review behavior feature vector; An intelligent value analysis module is used to analyze the implicit workload of the review behavior feature vector, and generate a quality weight factor; A remuneration calculation module is used to acquire project environment parameters, and calculate a remuneration amount according to the project environment parameters, the quality weight factor, and a preset settlement coefficient; A self-checking payment module is used to perform an abnormality check on the remuneration amount, and if the check is passed, a payment operation is performed, otherwise an artificial review process is triggered; A feedback optimization module is used to update the settlement coefficient according to the feedback data after payment.

[0051] It should be noted that the above formulae can be converted into unitless standard values or parameters of the same dimension that can be superimposed by using the dimensional consistency principle and mathematical standardization methods (for example, normalization processing, dimensionless parameter conversion, or unit system unification), so as to eliminate the interference of different dimensions on the operation logic, make the formula have mathematical operation rationality and objective law adaptability while preserving the original data distribution characteristics. It is a conventional technical means, and will not be described here. The electrical connection between the above-mentioned units does not necessarily mean direct connection, and indirect connection can also be used as long as the purpose of the application is achieved. The above-mentioned is only an exemplary embodiment of the application, and cannot limit the scope of the application.

[0052] That is, any equivalent changes and modifications made according to the teachings of the present application are still within the scope of the present application. Other embodiments of the present application will be readily apparent to those skilled in the art upon considering the description and practice of the principles disclosed herein. The present application is intended to cover any variations, uses, or adaptive changes of the present application that follow the general principles of the present application and include common knowledge or conventional technical means in the art that are not described in the present application.

Claims

1. An intelligent processing method for automated settlement of evaluation expert evaluation service remuneration, characterized in that: The method comprises: Obtain the text revision trajectory data and collaborative interaction data of review experts to generate review behavior feature vectors; Performing implicit workload analysis on the review behavior feature vector to generate a quality weight factor; Obtaining project environmental parameters, and calculating the remuneration amount based on the project environmental parameters, the quality weight factor, and a preset settlement coefficient; The remuneration amount is checked for abnormalities. If the check passes, the payment operation is executed; otherwise, a manual review process is triggered; The settlement coefficient is updated according to the feedback data after payment.

2. The intelligent processing method for automated settlement of evaluation expert evaluation service remuneration according to claim 1 is characterized in that: Generating the review behavior feature vector includes: Collect the text revision trajectory data of the review experts on the review documents and extract the number of revision points per thousand words as the revision density feature; Capture the collaborative interaction data of review experts and count the number of times the review suggestions are cited by other experts as the collaborative heat value; The revision density feature and the collaborative heat value are combined to generate a review behavior feature vector.

3. The intelligent processing method for automated settlement of evaluation expert evaluation service remuneration according to claim 2 is characterized in that: Generating the quality weight factor includes: Calculate a basic quality score based on the revised density characteristics; Calculate the synergistic contribution points according to the synergistic thermal value; The quality weight factor is calculated using the basic quality score, the collaborative contribution score and the preset weight ratio.

4. The intelligent processing method for automated settlement of evaluation expert evaluation service remuneration according to claim 3 is characterized in that: The calculated remuneration amount includes: Obtaining project environment parameters, and screening review items for which the number of experts with different opinions is greater than a preset threshold based on the project environment parameters, to construct a collection of controversial projects; Calculate the additional dispute correction coefficient based on the disputed item set; When a cross-border payment identifier is detected, the real-time exchange rate fluctuation buffer pool is called to calculate the exchange rate compensation and obtain the exchange rate compensation value; The amount of remuneration is calculated using the additional dispute correction coefficient, the exchange rate compensation value, the quality weight factor and a preset settlement coefficient.

5. The intelligent processing method for automatic settlement of evaluation expert evaluation service remuneration according to claim 4 is characterized in that: The method further comprises: Obtaining a project deadline and a current time according to the project environment parameters, and calculating a difference between the project deadline and the current time to obtain a project deadline time value; Using the project deadline time value, calculate the project urgency coefficient; Obtaining a review operation timestamp, and generating a cross-time zone review identifier when the review operation timestamp is in a preset nighttime time period; When the cross-time zone review identifier is identified, the quality weight factor is modified according to the project urgency coefficient.

6. The intelligent processing method for automated settlement of evaluation expert evaluation service remuneration according to claim 4 is characterized in that: The abnormality check of the remuneration amount includes: Obtain historical review behavior data and corresponding historical remuneration of review experts, wherein the historical review behavior data includes historical review start time, historical end time, and historical number of reviewed documents; Calculate the historical review efficiency using the number of historical review documents; Construct a mapping table between historical review efficiency and corresponding historical remuneration; Obtain the total number of review documents completed by the review expert within the preset time window from the current time point forward, and calculate the current review efficiency; Obtaining a remuneration benchmark value according to the current review efficiency and the relationship mapping table; Calculate the difference between the remuneration amount and the remuneration benchmark value; When the reward difference is greater than a preset reward threshold, an abnormality check flag is generated.

7. The intelligent processing method for automatic settlement of evaluation expert evaluation service remuneration according to claim 6 is characterized in that: The triggering manual review process includes: When the abnormal check mark exists, a manual review process is triggered; Conducting a discrepancy analysis between the remuneration benchmark value and the remuneration amount; The weight ratio is adjusted according to the difference analysis results.

8. The intelligent processing method for automated settlement of evaluation expert evaluation service remuneration according to claim 1 is characterized in that: The updating of the settlement coefficient according to the feedback data after payment includes: Obtaining feedback from review experts on their satisfaction with the remuneration amount, obtaining satisfaction scores and storing them in a satisfaction database; Determining whether an update is needed based on the satisfaction database; When an update is required, the satisfaction coefficient is calculated using the data in the satisfaction database; Calculating a settlement correction factor based on the satisfaction coefficient and the quality weight factor; The settlement coefficient is updated using the settlement correction factor.

9. The intelligent processing method for automated settlement of evaluation expert evaluation service remuneration according to claim 1 is characterized in that: The method further comprises: After the payment operation is completed, the expert behavior change data within the preset period is continuously collected. The expert behavior change data includes the response speed of subsequent review projects, the acceptance rate of review documents, and the frequency of collaborative interaction; Quantifying the expert behavior change data and calculating a comprehensive participation index; When the comprehensive participation index is lower than a preset baseline, the collection dimension of the review behavior feature vector is updated.

10. An intelligent processing system for automated settlement of review expert review remuneration, applied to the intelligent processing method for automated settlement of review expert review remuneration as described in any one of claims 1 to 9, characterized in that: The system comprises: The review behavior collection module is used to obtain the text revision trajectory data and collaborative interaction data of review experts and generate the review behavior feature vector; An intelligent value analysis module, configured to perform implicit workload analysis on the review behavior feature vector and generate a quality weight factor; A remuneration calculation module, configured to obtain project environmental parameters and calculate a remuneration amount based on the project environmental parameters, the quality weight factor, and a preset settlement coefficient; A self-checking payment module is used to check the remuneration amount for abnormalities. If the check passes, the payment operation is executed; otherwise, a manual review process is triggered; A feedback optimization module is used to update the settlement coefficient according to the feedback data after payment.

Images