Task scheduling method and related product

Abstract

The application discloses a task scheduling method and related products, the method comprises the following steps: obtaining the target risk level corresponding to the target task to be allocated; then adjusting the initial matching parameter set according to the target risk level to obtain the target matching parameter set; then screening the initial object set according to the target matching parameter set to obtain the first object set; further screening the first object set according to the load quota threshold to obtain the second object set; finally, based on the preset task allocation strategy, the target task is allocated to the target object. In this way, the occurrence probability of the problem of mismatching between high-risk tasks and ordinary audit objects and unbalanced allocation of human resources is reduced, and the processing efficiency and reliability of task scheduling are further improved.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a task scheduling method and related products. Background Technology

[0002] Task scheduling refers to the process of assigning multiple tasks to different business personnel according to their needs. With the development of computer technology, task scheduling has gradually changed from manual allocation to automatic allocation, which has improved the efficiency of task scheduling.

[0003] In related technologies, task scheduling is generally achieved through static matching, which assigns multiple tasks to different business personnel based on pre-defined matching rules (such as job position, permissions, or task limits). However, static matching can lead to unreasonable task allocation; for example, a task may not match the business personnel's expertise, or multiple complex tasks may be assigned to the same business personnel. This can reduce the efficiency and reliability of task scheduling. Summary of the Invention

[0004] To address the aforementioned issues, this application provides a task scheduling method and related products, aiming to improve the processing efficiency and reliability of task scheduling.

[0005] The embodiments of this application disclose the following technical solutions: In a first aspect, embodiments of this application provide a task scheduling method, which may include: Obtain the target risk level corresponding to the target task to be assigned; The initial matching parameter set is adjusted according to the target risk level to obtain the target matching parameter set. The initial matching parameter set includes at least one of the following initial matching parameters: business scenario parameters, task object parameters, review permission parameters, or professional capability parameters. The initial object set is filtered according to the target matching parameter set to obtain the first object set. The initial object set includes multiple initial objects, and the first object set includes at least one first object, which is an initial object that matches the target matching parameter set. The first set of objects is filtered according to the load quota threshold to obtain the second set of objects. The second set of objects includes at least one second object, which is a first object that meets the load quota threshold. The load quota threshold represents the upper limit of the load quota corresponding to the first object. Based on a preset task allocation strategy, the target task is assigned to the target object, which is a second object in a second object set.

[0006] In conjunction with the first aspect, in one possible implementation, the target risk level corresponding to the target task to be assigned is obtained, including: Obtain the target task information corresponding to the target task; Feature extraction processing is performed on the target task information to obtain target risk features; The target risk level is determined based on the target risk characteristics and the preset risk level rules.

[0007] In conjunction with the first aspect, in one possible implementation, the initial matching parameter set is adjusted according to the target risk level to obtain the target matching parameter set, which includes: Determine the target weight coefficients for each initial matching parameter in the initial matching parameter set based on the target risk level; The target matching parameter set is determined based on each initial matching parameter and its corresponding target weight coefficient.

[0008] In conjunction with the first aspect, in one possible implementation, the first set of objects is filtered according to a load quota threshold to obtain a second set of objects, including: Obtain the basic load quota value corresponding to each first object. The basic load quota value includes the individual load quota value and the team load quota value. The individual load quota value is determined based on at least one of the audit duration, audit risk, or audit cost. The team load quota value represents the sum of the basic load quota values ​​of all corresponding objects. Select the first object from the first object set that meets the negative basic load quota condition to obtain the second object set; the basic load quota condition is that the individual load quota value is less than the first basic load quota value, and the team load quota value is less than the second basic load quota value.

[0009] In conjunction with the first aspect, in one possible implementation, the first set of objects is filtered according to a load quota threshold to obtain a second set of objects, including: Obtain the risk load quota value corresponding to each first object. The risk load quota value is determined based on at least one of the audit limit or audit risk. Select the first object from the first object set that meets the risk load quota condition to obtain the second object set; the risk load quota condition is that the risk load quota value is less than the risk load quota value.

[0010] In conjunction with the first aspect, in one possible implementation, the target task is assigned to the target object based on a preset task allocation strategy, including: If the preset task allocation strategy is automatic allocation mode, the target task will be assigned to the highest priority second object in the second object set; If the preset task allocation strategy is preemptive allocation mode, then the target task will be allocated to the second object in the second object set that triggered the preemption operation; If the preset task allocation strategy is manual allocation mode, the target task will be allocated to the second object in the second object set that triggered the specified operation.

[0011] In conjunction with the first aspect, one possible implementation also includes determining the professional competence parameters corresponding to the initial object in the following way: Obtain the set of historical profile factors corresponding to the initial object. The set of historical profile factors includes at least one of the following: historical task type, historical error rate, historical average processing time, or historical processing permissions. By performing a weighted calculation on all historical portrait factors in the historical portrait factor set, the professional ability parameters corresponding to the initial object are obtained.

[0012] Secondly, embodiments of this application provide a task scheduling device, which may include: The acquisition unit is used to acquire the target risk level corresponding to the target task to be assigned. The adjustment unit is used to adjust the initial matching parameter set according to the target risk level to obtain the target matching parameter set. The initial matching parameter set includes at least one of the following initial matching parameters: business scenario parameters, task object parameters, review authority parameters, or professional capability parameters. The first filtering unit is used to filter the initial object set according to the target matching parameter set to obtain the first object set. The initial object set includes multiple initial objects, and the first object set includes at least one first object, which is an initial object that matches the target matching parameter set. The second filtering unit is used to filter the first object set according to the load quota threshold to obtain a second object set. The second object set includes at least one second object, which is a first object that meets the load quota threshold. The load quota threshold represents the upper limit of the load quota corresponding to the first object. The task scheduling unit is used to assign target tasks to target objects based on a preset task allocation strategy. The target object is a second object in a second object set.

[0013] 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 perform the task scheduling method as described in the first aspect.

[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which is loaded by a processor to execute the task scheduling method as described in the first aspect.

[0015] Beneficial effects: The task scheduling method provided in this application involves obtaining the target risk level corresponding to the target task to be assigned; then adjusting the initial matching parameter set according to the target risk level to obtain a target matching parameter set; then filtering the initial object set according to the target matching parameter set to obtain a first object set; further filtering the first object set according to the load quota threshold to obtain a second object set; and finally assigning the target task to the target object based on a preset task allocation strategy.

[0016] The initial matching parameter set includes at least one of the following: business scenario parameters, task object parameters, audit permission parameters, or professional capability parameters; the initial object set includes multiple initial objects, the first object set includes at least one first object, which is an initial object that meets the target matching parameter set; the second object set includes at least one second object, which is a first object that meets the load quota threshold, and the load quota threshold represents the upper limit of the load quota corresponding to the first object; the target object is a second object in the second object set.

[0017] In this way, by dynamically adjusting the initial matching parameter set, which includes at least one parameter from the business scenario, task object, review authority, or professional capability parameters, based on the target risk level corresponding to the target task, a target matching parameter set is obtained. Then, the initial object set is initially screened based on the target matching parameter set, and a second screening is carried out in combination with the load quota threshold. Finally, the task is dispatched based on the preset allocation strategy, realizing the matching rules for task scheduling based on risk level differentiation. It also comprehensively considers the task processing capabilities of the review object from multiple dimensions, and balances the allocation of tasks in combination with the real-time load status of the review object, thereby reducing the probability of mismatch between high-risk tasks and ordinary review objects and imbalance in human resource allocation, and further improving the processing efficiency and reliability of task scheduling. Attached Figure Description

[0018] 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.

[0019] Figure 1 A flowchart illustrating a task scheduling method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of a task scheduling device provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a control device provided in an embodiment of this application. Detailed Implementation

[0020] As described earlier, taking expense reimbursement as an example, the expense reimbursement process involves the full lifecycle review of reimbursement documents for employee travel, material procurement, labor settlement, and business entertainment. The current manual offline registration and random allocation of documents is insufficient to handle the massive volume of documents. Therefore, related technologies provide a static matching method for expense reimbursement, which assigns different types of reimbursement documents to different business personnel based on pre-defined matching rules.

[0021] However, the static matching methods provided in related technologies may not be suitable for the refined, compliant, or efficient auditing needs of enterprises, resulting in at least one of the following technical defects: First, the matching dimension is singular, relying solely on job permissions for coarse matching, without differentiating personnel's professional capabilities. Traditional dispatching models only rely on basic permissions for fixed positions to complete coarse-grained work order matching, lacking a differentiated screening and matching mechanism for on-duty personnel within the same review role. The system cannot accumulate or reuse historical work data of reviewers, and cannot accurately distinguish each person's expertise in reimbursement business types, historical document rejection error rates, processing time for large / high-risk documents, or qualifications for specialized reimbursement business such as travel / purchasing / labor services. This leads to the random allocation of large, special, and high-risk business documents, frequently resulting in mismatches where ordinary reviewers undertake high-risk documents while senior professionals handle small, simple documents, directly causing insufficient review professionalism, missed financial risk assessments, and persistently high document rework rates.

[0022] Second, the fixed weights of the matching rules prevent dynamic adjustment of dimension priorities based on document risk, resulting in weak scenario adaptability. Traditional dispatch rules use fixed parameters for the weights of each matching dimension, applying the same matching weight logic to both low-risk, small-amount reimbursement and large-amount, abnormal, high-risk documents. Redundant verification of job and service recipient information slows down overall batch dispatch efficiency for low-risk standardized documents; while weakened verification of job responsibilities and personnel professionalism for high-risk specialized documents easily leads to mismatches of responsibilities and documents, failing to adapt to the differentiated business scenarios of standardized batch review and high-risk specialized refined review, exhibiting extremely poor scenario adaptability.

[0023] Third, the load limiting mechanism is crude, and load statistics are severely distorted, with gaps in both individual and team-level limits. The existing work order system simply uses the number of documents as the statistical basis, setting a uniform task limit for each individual, without differentiating the review time and cost, or risk verification cost, for different documents. The workload and risk cost of reviewing a single large, abnormal document far exceed those of multiple regular, small documents. A uniform counting threshold easily leads to a load imbalance problem where "small documents are piled up by highly capable reviewers, while high-risk documents go unhandled." Furthermore, the current technology only limits the workload of individual workers, without setting an overall team load threshold or a cross-position manpower adjustment mechanism. When a single position is fully loaded, documents continuously accumulate, and there is no dynamic manpower replacement channel, resulting in extremely low overall manpower utilization.

[0024] Fourth, the methods for identifying document risks are inadequate, and documents with hidden risks are easily mistakenly included in the automated order dispatch process. Traditional solutions rely solely on structured, fixed fields in expense reimbursement forms to categorize documents, failing to parse unstructured text attachments such as invoice images, related business contracts, reimbursement remarks, and travel instructions. They cannot automatically extract hidden risk characteristics such as excessive amounts, third-party related expenditures, temporary special approvals, and abnormal spending, and cannot accurately classify document risk levels in advance. A large number of documents with hidden compliance risks are incorrectly assigned to the automated batch order dispatch process, lacking a dedicated manual review step, which greatly increases the risk of financial compliance problems for enterprises.

[0025] Fifth, the dispatching mode is rigid and singular, lacking adaptive switching logic and incomplete scenario coverage. Traditional dispatching systems only support a single automatic dispatching mode, without configuring order grabbing or manual special dispatching channels. This makes it impossible to specifically adapt to three differentiated operational scenarios: batch routine documents, cross-functional general documents, and complex and high-risk documents. It cannot achieve unattended batch efficiency improvements, nor can it handle dynamic manpower replenishment or targeted special review of high-risk documents, ultimately leading to a continuous backlog of work orders and low overall review efficiency.

[0026] In summary, related technologies generally employ static matching for task scheduling, assigning multiple tasks to different business personnel based on pre-defined matching rules (such as job position, permissions, or task limits). However, static matching can lead to unreasonable task allocation; for example, a task might not match the expertise of a business personnel, or multiple complex tasks might be assigned to the same person. This can reduce the efficiency and reliability of task scheduling.

[0027] Based on this, embodiments of this application provide a task scheduling method and related products. The method includes: obtaining the target risk level corresponding to the target task to be assigned; then adjusting the initial matching parameter set according to the target risk level to obtain a target matching parameter set; then filtering the initial object set according to the target matching parameter set to obtain a first object set; further filtering the first object set according to the load quota threshold to obtain a second object set; and finally assigning the target task to the target object based on a preset task allocation strategy.

[0028] The initial matching parameter set includes at least one of the following: business scenario parameters, task object parameters, audit permission parameters, or professional capability parameters; the initial object set includes multiple initial objects, the first object set includes at least one first object, which is an initial object that meets the target matching parameter set; the second object set includes at least one second object, which is a first object that meets the load quota threshold, and the load quota threshold represents the upper limit of the load quota corresponding to the first object; the target object is a second object in the second object set.

[0029] In this way, by dynamically adjusting the initial matching parameter set, which includes at least one parameter from the business scenario, task object, review authority, or professional capability parameters, based on the target risk level corresponding to the target task, a target matching parameter set is obtained. Then, the initial object set is initially screened based on the target matching parameter set, and a second screening is carried out in combination with the load quota threshold. Finally, the task is dispatched based on the preset allocation strategy, realizing the matching rules for task scheduling based on risk level differentiation. It also comprehensively considers the task processing capabilities of the review object from multiple dimensions, and balances the allocation of tasks in combination with the real-time load status of the review object, thereby reducing the probability of mismatch between high-risk tasks and ordinary review objects and imbalance in human resource allocation, and further improving the processing efficiency and reliability of task scheduling.

[0030] 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.

[0031] 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.

[0032] See Figure 1 The figure is a flowchart illustrating a task scheduling method provided in an embodiment of this application.

[0033] Combination Figure 1 As shown in the embodiments of this application, the task scheduling method may include: S11: Obtain the target risk level corresponding to the target task to be assigned.

[0034] The target task refers to a task to be assigned. For example, a target task can be a reimbursement work order in a company's financial expense reimbursement process that has reached the dispatch node but has not yet been assigned to a reviewer. This work order integrates all business attachment information, including structured form data, invoice images, business contracts, reimbursement remarks, and travel instructions. As an example, a target task can be, but is not limited to, small-amount travel reimbursement work orders, large-amount purchase reimbursement work orders, external settlement work orders, and abnormal consumption reimbursement work orders, etc., without specific limitations.

[0035] The target risk level refers to the level of risk present in the target task. The target risk level can be used to quantitatively characterize the relative degree of potential risk of a single expense reimbursement work order. For example, the target risk level can be, but is not limited to, low risk level, medium risk level, high risk level, etc., without specific limitations here.

[0036] It should be understood that by obtaining the target risk level corresponding to the target task to be assigned, this step can adjust the subsequent matching parameters through pre-risk verification. To a certain extent, this can reduce the probability of tasks with compliance risks being directly assigned to mismatched business personnel and improve the reliability of task scheduling.

[0037] S12: Adjust the initial matching parameter set according to the target risk level to obtain the target matching parameter set.

[0038] The initial matching parameter set refers to the original set of multi-dimensional matching parameters that are pre-configured and have not been dynamically corrected. The initial matching parameter set can be used to store all the basic judgment dimensions required to screen qualified audit objects. In the embodiments of this application, the initial matching parameter set may include at least one of the following initial matching parameters: business scenario parameters, task object parameters, audit permission parameters, or professional capability parameters.

[0039] Business scenario parameters refer to quantitative identifiers that characterize the business category to which the target task belongs. Business scenario parameters can be used to distinguish differentiated business types such as business travel, material procurement, external settlement, business entertainment, and special project expenditures. For example, business scenario parameters can be quantitative identifiers for business travel scenarios, large-scale project procurement scenarios, and external labor settlement scenarios, etc., without specific limitations.

[0040] The task object parameter refers to the quantitative parameters that characterize the service entity, external partner, or internal responsible department corresponding to the expenditure of the target task. The task object is used to verify whether the candidate review object has the special review and processing authority for the corresponding service entity or partner's documents. For example, the task object parameter can be the internal branch department identifier, the external third-party partner enterprise identifier, the off-site special project entity identifier, etc., without specific limitations here.

[0041] The review authority parameter refers to the quantitative parameters of authority formed based on the enterprise's organizational structure and the division of job responsibilities. The review authority parameter can be used to initially determine whether the candidate reviewer has the basic order acceptance qualification for the corresponding position and the corresponding amount level. For example, the review authority parameter can be the authority identifier for the basic financial initial review position, the authority identifier for the group's large-amount special review position, the exclusive review authority identifier for regional branches, etc., without specific limitations here.

[0042] Professional competence parameters refer to the quantitative scores of capabilities obtained based on the historical operational data of the audited entity. These parameters can be used to characterize an individual's proficiency in handling expense reports for specific business types, their level of risk and error control, and their efficiency in processing complex documents. For example, professional competence parameters could be scores for travel business matching, scores for handling risks in large-scale procurement, and scores for error control in labor service documents, etc., without further specific limitations.

[0043] The target matching parameter set refers to the new matching parameter set obtained by adjusting the sub-parameters within the initial matching parameter set according to the target risk level. The weights of each dimension parameter in the target matching set change synchronously with the risk level of the document, serving as a dynamic judgment standard for selecting suitable audit objects.

[0044] As an example, the weight coefficients of each dimension can be dynamically calculated in real time based on the target risk level output in step S11 to achieve differentiated and precise matching: for low-risk ordinary reimbursement documents, the weight of business scenario parameters can be increased, while the weight of audit authority parameters and professional ability parameters can be decreased, so as to improve the batch dispatch efficiency of a large number of regular documents; for large-amount documents or abnormally high-risk documents, the task object parameters, audit authority parameters, and professional ability parameters can be increased to strengthen the matching of audit authority and responsibility and the verification of personnel professional ability, thereby reducing the probability of audit errors caused by mismatch of high-risk documents.

[0045] It should be understood that this step reconstructs a target matching parameter set that adapts to the risk level corresponding to the target task by dynamically adjusting the weight coefficients of each initial match in the initial matching parameter set according to the target risk level. This achieves a differentiated screening mechanism for different tasks, taking into account both task processing efficiency and compliance requirements, and can improve the processing efficiency and reliability of task scheduling to a certain extent.

[0046] S13: Filter the initial object set according to the target matching parameter set to obtain the first object set.

[0047] The initial object set refers to the collection of all reviewers who are capable of accepting tasks. In this embodiment of the application, the initial object set includes multiple initial objects (i.e., all reviewers).

[0048] The first object set refers to the subset of reviewers obtained after the initial filtering of the initial object set using the target matching parameter set as the screening criterion. The first object set includes at least one first object, which is an initial object that meets the target matching parameter set.

[0049] It should be understood that this step is the initial screening stage, which uses the target matching parameter set dynamically generated in step S12 as a unified verification standard to perform multi-dimensional matching and filtering on all initial objects with permissions, thereby obtaining the first set of objects that meet all conditions, realizing multi-dimensional weighted comprehensive matching; at the same time, professional ability parameters are included in the screening benchmark to complete the initial adaptation of document business attributes, risk level and reviewer professional qualifications, filtering out a large number of candidates with mismatched business and insufficient professional ability, which can improve the processing efficiency and reliability of task scheduling to a certain extent.

[0050] S14: Filter the first object set according to the load quota threshold to obtain the second object set.

[0051] The load quota threshold refers to the quantitative standard of the upper limit of task load. In the embodiments of this application, the load quota threshold can be divided into individual load quota threshold and team load quota threshold. The load quota threshold represents the upper limit of the load quota corresponding to the first object. The load quota threshold can be used to constrain the maximum total work order load that a single person can undertake and the total total load that all personnel in a single position can undertake. Subsequently, the load quota threshold can be used as a quantitative boundary to determine whether the reviewer has the conditions to accept new orders.

[0052] Load quota refers to a standardized load measurement value calculated for a single task. Load quota can be used to quantitatively represent the human resources occupied by a document. For example, one small travel document corresponds to one basic load quota, and one million-level purchase document corresponds to two risk load quotas, etc. No specific limitation is made here.

[0053] The second set of objects refers to the final subset of candidates obtained after conducting load compliance verification on personnel who have completed business adaptation screening within the first set, using load quota thresholds as the verification standard. All second objects in the set simultaneously meet the triple constraints of multi-dimensional business matching conditions, individual load not exceeding limits, and the load of their respective teams not exceeding limits, so as to serve as a pool of selectable personnel for the work order allocation process.

[0054] It should be understood that this step performs a secondary screening of all first objects in the first object set based on the load quota threshold, filtering out first objects whose individual load exceeds the limit or whose team load is full, and outputting a second object set that matches the business qualifications and has available workload. This addresses, to some extent, the technical shortcomings of traditional order dispatch systems, such as distorted load statistics, limiting the number of tasks for a single person, and lacking team load and cross-position adjustment mechanisms. By differentiating the manpower costs of different documents through differentiated load quotas, it accurately identifies the actual workload saturation of personnel. At the same time, it uses team load thresholds to achieve cross-position manpower adjustment screening, balancing the differences in busyness and idleness among personnel within the review team, reducing the phenomenon of local backlog of work orders, and improving the processing efficiency and reliability of business scheduling to a certain extent.

[0055] S15: Assign the target task to the target object based on the preset task allocation strategy.

[0056] The target object is a second object within the second object set. The target object refers to a single reviewer selected from all second objects within the second object set, according to the currently enabled preset task allocation strategy, who will be the final recipient of the current target task.

[0057] The preset task allocation strategy refers to the pre-configured assignment execution logic, which is used to standardize the work order selection rules for candidates within the second object set. For example, there are automatic allocation strategies for low-risk documents, preemptive allocation strategies for medium-risk documents, and manual allocation strategies for high-risk documents, etc., which are not specifically limited here.

[0058] It should be understood that, based on the determination of the second set of objects, this step selects a suitable second object from the second set of objects as the target object according to the preset task allocation strategy. This can adaptively adjust the task scheduling method, thereby making the task scheduling result more in line with the user's actual needs and improving the reliability of task scheduling to a certain extent.

[0059] It should be understood that the task scheduling method provided in this application dynamically adjusts the initial matching parameter set, which includes at least one parameter among business scenario, task object, review authority, or professional capability parameters, based on the target risk level corresponding to the target task, to obtain the target matching parameter set. Then, the initial object set can be initially screened based on the target matching parameter set, followed by a secondary screening based on the load quota threshold. Finally, the task is dispatched based on a preset allocation strategy. This achieves the matching rules for task scheduling based on risk level differentiation, comprehensively considers the task processing capabilities of the review object from multiple dimensions, and balances the allocation of tasks based on the real-time load status of the review object. This reduces the probability of mismatch between high-risk tasks and ordinary review objects, and imbalance in human resource allocation, thereby further improving the processing efficiency and reliability of task scheduling.

[0060] Based on the task scheduling method provided in the above embodiments, in one possible implementation, step S11 may include: A1: Obtain the target task information corresponding to the target task.

[0061] The target task information refers to all structured and unstructured complete business data contained in the target task. For example, structured information may include, but is not limited to, fixed fields in forms such as the amount to be filled in, business type dropdown options, reporting department, and the applicant's position. Unstructured information may include, but is not limited to, text from invoice OCR images, scanned copies of business contracts, reimbursement notes, travel itinerary descriptions, and specially approved documents. The target task information is used to completely reconstruct the full-dimensional business details of the documents, providing a comprehensive data source for risk feature extraction.

[0062] A2: Perform feature extraction processing on the target task information to obtain target risk features.

[0063] Feature extraction refers to the process of extracting key features from target task information. For example, the target task information can first undergo a series of standardized computational processes, including keyword retrieval, numerical threshold comparison, correlation matching, and anomaly semantic recognition, from structured fields, image text, and attachment text. Then, the standardized target task information is processed using a pre-trained feature extraction model to obtain target risk features.

[0064] The target risk feature refers to the standardized abnormal signal quantification identifier output after feature extraction processing. It is used to characterize various risk anomalies in the target task and serves as the basic input factor for risk level determination. For example, the target risk feature may include, but is not limited to, numerical risk features, text-related risk features, and missing attachment risk features, etc., without specific limitations here.

[0065] It should be understood that this step, by performing feature processing on the target task information, can identify the hidden risk characteristics in the target task, so as to facilitate the subsequent identification and classification of the risk level of the target task.

[0066] A3: Determine the target risk level corresponding to the target task based on the target risk characteristics and the preset risk level rules.

[0067] Preset risk level rules refer to pre-configured quantitative grading judgment standards. The rules define the risk scores corresponding to various target risk characteristics and the risk levels matched by different total score intervals. They are used to transform discrete risk characteristics into standardized low, medium and high risk stratification results. For example, a single minor abnormal feature corresponds to low risk, multiple superimposed abnormal features correspond to medium risk, and a large amount of money superimposed with multiple abnormal features corresponds to high risk, etc. No specific limitations are made here.

[0068] It should be understood that this step, through the three-step linkage of task information collection, automatic extraction of multi-dimensional risk features, and standardized rule-based risk grading, achieves the automated determination of the target risk level corresponding to the target task. To a certain extent, this compensates for the technical deficiencies of traditional dispatch systems, such as limited data sources for risk identification and inability to parse hidden risks in unstructured attachments. It expands the coverage of abnormal feature identification, unifies risk stratification and judgment standards, reduces the manpower consumption of manually marking document risks in advance, and lowers the probability of processing errors caused by task allocation errors that hide compliance risks.

[0069] Based on the task scheduling method provided in the above embodiments, in one possible implementation, step S12 may include: B1: Determine the target weight coefficients for each initial matching parameter in the initial matching parameter set based on the target risk level.

[0070] The target weight coefficient refers to the weight percentage determined for at least one of the initial matching parameters, namely business scenario parameters, task object parameters, review authority parameters, or professional capability parameters. The higher the target weight coefficient, the more important the initial matching parameter is, and the higher the threshold for subsequent successful matching.

[0071] B2: Determine the target matching parameter set based on each initial matching parameter and its corresponding target weight coefficient.

[0072] It should be understood that this embodiment can allocate target weight coefficients for each matching dimension based on the target risk level. For example, taking expense reimbursement documents as an example, the weight coefficients of the review authority parameter and professional ability parameter are weakened for low-risk routine expense reimbursement documents to simplify the screening calculation logic, shorten the personnel screening time for a large number of documents, and improve the efficiency of batch dispatch processing. For high-risk and difficult expense reimbursement documents, the weight coefficients of the task object parameter, review authority parameter, and professional ability parameter can be strengthened to increase the proportion of compliance and professional ability indicators in the calculation of personnel comprehensive matching scores. To a certain extent, this increases the probability of matching high-risk documents with senior professional reviewers and reduces the phenomenon of mismatched reviewers for high-risk documents. In this way, the processing efficiency and reliability of task scheduling can be improved to a certain extent.

[0073] In one possible implementation of the task scheduling method provided in the above embodiments, step S14 may include: C1: Get the base load quota value corresponding to each first object.

[0074] The basic load quota value refers to the standardized load measurement data value calculated for a target task that is a routine task. For example, assuming the target task is a routine small-amount standardized expense reimbursement document, the basic load quota value can be a standardized load measurement value calculated based on the average document review time and basic verification cost. In this embodiment, the basic load quota value can be divided into an individual basic load quota value and a team basic load quota value. The individual basic load quota value represents the total cumulative load of routine documents currently undertaken by a single person, and the team load quota value represents the total load of routine documents for all personnel in a single position. The basic load quota value is used to limit the upper limit of manpower allocation for routine batch documents.

[0075] It should be understood that individual workload quotas are determined based on at least one of the following: review duration, review risk, or review cost. Review duration refers to the review time for a single task; review risk refers to the risk level of a single task; and review cost refers to the human or time cost required to review a single task. For example, assuming document A has a review duration of 1 minute, a low review risk, and a review cost of x1, then document A can be allocated 1 workload quota. If document B has a review duration of 20 minutes and a high review risk, then document B can be allocated 2 workload quotas. Correspondingly, if the reviewer includes both document A and document B, then their individual workload quota value can be 3.

[0076] It should be understood that the team load quota value represents the sum of the base load quota values ​​of all corresponding objects. For example, if a team includes 3 auditors, then the sum of the base load quota values ​​of these 3 auditors is the team load quota value.

[0077] C2: Select the first object from the first object set that meets the negative base load quota condition, and obtain the second object set.

[0078] The basic load quota condition refers to the dual quantification constraints that must be simultaneously met for a first object in a first object set to be included in a second object set; if either condition is not met, the first object will fail the basic load check. In this embodiment, the basic load quota condition is that the individual load quota value is less than the first basic load quota value, and the team load quota value is less than the second basic load quota value.

[0079] The first basic load quota value refers to the upper limit threshold of the basic load quota that a single person can undertake for regular and standardized tasks, serving as a quantitative boundary for determining whether a person can add new tasks.

[0080] The second basic load quota value refers to the total upper limit threshold of the basic load quota corresponding to the standard and routine tasks that all personnel in a single position can bear, serving as the boundary for determining the overall task-bearing capacity of the position.

[0081] It should be understood that the embodiments of this application calculate the basic load quota value based on at least one parameter among the audit time, audit risk or audit cost of the target task, replacing the traditional coarse-grained measurement method that only counts the number of tasks. This can objectively reflect the degree to which the actual number of audit personnel is occupied by the task, and to a certain extent improve the technical shortcomings of the traditional load statistics method that cannot distinguish the differences in the workload of documents and the distortion of statistical results, and accurately determine the true saturation level of personnel for regular documents.

[0082] At the same time, by setting dual constraints of individual basic load thresholds and team basic load thresholds, not only can the total amount of tasks that an individual can undertake be limited, but also the overall task capacity limit of a single position can be limited. When a position is fully loaded, it is restricted from further assigning tasks to personnel in that position. It can also be linked to a cross-position adjustment mechanism to expand the scope of candidates, which can alleviate the business problems of continuous backlog of regular batch tasks in a single position and uneven utilization of manpower in a single position to a certain extent.

[0083] In one possible implementation of the task scheduling method provided in the above embodiments, step S14 may include: D1: Obtain the risk load quota value corresponding to each first object.

[0084] The risk load quota value is determined based on at least one of the audit limit or audit risk. The audit limit refers to the amount of data for a single task; for example, the audit limit for an expense report represents the amount of reimbursement.

[0085] Risk load quota refers to a standardized load measurement value calculated based on the audit limit or audit risk for large, abnormal, or high-risk tasks. It is used to quantify the additional human review resources required for high-risk tasks. For example, a multi-abnormal purchase order with a value of one million corresponds to three risk load quotas.

[0086] D2: Select the first object from the first object set that meets the risk load quota conditions to obtain the second object set.

[0087] The risk load quota condition refers to the quantitative constraint that the first object must meet to undertake new high-risk tasks. That is, the risk load quota value corresponding to the first object must be less than the risk load quota value before it can be included in the second object set; otherwise, the first object that does not meet this condition will not be included in the second object set that can undertake high-risk tasks.

[0088] It should be understood that, in this embodiment of the application, the risk load quota value corresponding to the first object is quantified by at least one of the audit amount or audit risk of the target task, and then each first object in the first object set is screened by the risk load threshold to finally obtain the second object set. This accurately quantifies the risk cost of large and difficult tasks and reduces the probability of high-risk tasks being backlogged and audit omissions occurring.

[0089] In one possible implementation of the task scheduling method provided in the above embodiments, step S15 may include: E1: If the preset task allocation strategy is automatic allocation mode, the target task will be assigned to the second object with the highest priority in the second object set.

[0090] Automatic allocation mode refers to a mode in which target tasks are automatically allocated by pre-set allocation logic. Automatic allocation mode requires no manual intervention. For example, it can automatically sort the second objects in the second object set from high to low according to their remaining available load capacity, and then select the second object at the top of the sorted list to undertake the target task.

[0091] E2: If the preset task allocation strategy is preemptive allocation mode, then the target task will be allocated to the second object in the second object set that triggered the preemption operation.

[0092] The preemptive allocation mode refers to an allocation mode in which any one of the second objects in the second object set actively accepts the task. For example, a target task can be pushed to a shared task pool, allowing only second objects within the second object set to trigger the order-accepting preemptive operation independently. The target task can then be allocated to the first second object to complete the preemptive operation.

[0093] E3: If the preset task allocation strategy is manual allocation mode, the target task will be allocated to the second object in the second object set that triggered the specified operation.

[0094] Manual allocation mode refers to a mode in which allocation is performed manually. For example, automatic and preemptive allocation channels can be turned off, allowing administrators to visually view the second objects within the second object set and, based on parameters such as the professional matching score and real-time load status of the second objects, designate one of the second objects to receive the target object.

[0095] It should be understood that the embodiments of this application employ three independent assignment execution logics—automatic allocation mode, preemptive allocation mode, and manual allocation mode—which are individually linked to the low, medium, and high risk levels of tasks and adaptively switch between them. This, to a certain extent, compensates for the rigid and singular technical shortcomings of traditional task dispatching modes. Specifically, low-risk tasks can rely on automatic allocation to reduce manual intervention costs and improve batch processing efficiency; medium-risk tasks can rely on preemptive allocation to achieve cross-position manpower adjustment and smooth backlog of work orders; and high-risk tasks can rely on manual allocation to achieve targeted professional review and strengthen corporate financial compliance management.

[0096] Based on the task scheduling method provided in the above embodiments, in one possible implementation, the method further includes determining the professional capability parameters corresponding to the initial object in the following way: F1: Get the set of historical profile factors corresponding to the initial object.

[0097] The set of historical profile factors includes at least one of the following: historical task type, historical error rate, historical average processing time, or historical processing authority.

[0098] The historical profile factor set refers to the data collection of quantitative indicators of the full-cycle historical operations of a single initial object. The historical profile factor set may include, but is not limited to, at least one of the following historical profile factors: historical task type, historical error rate, historical average processing time, or historical processing authority. The historical profile factor set can be used to quantitatively restore the initial object's ability to handle various types of business, risk management level, or operational efficiency.

[0099] Historical profile factors refer to quantitative indicators that represent the historical performance of an initial object in a single dimension. Different historical profile factors are used to objectively measure the ability of personnel to handle specific business from different dimensions. For example, the historical error rate factor is the percentage of similar documents rejected by personnel in the past, and the historical average processing time factor is the average time spent by personnel to process similar documents, etc., without specific limitations here.

[0100] F2: Perform a weighted calculation on all historical portrait factors in the historical portrait factor set to obtain the professional ability parameters corresponding to the initial object.

[0101] It should be understood that, by obtaining the historical profile factor set of the initial object and performing weighted calculation, the professional capability parameters corresponding to the initial object can be obtained. This allows the implicit business expertise or risk management capabilities of the initial object to be transformed into standardized and calculable professional capability parameters. To a certain extent, this compensates for the technical shortcomings of traditional systems that cannot quantitatively distinguish the professional capabilities of personnel in the same position and rely solely on coarse matching based on fixed position permissions. This provides quantitative data support for the precise matching of documents and personnel differences.

[0102] Based on the task scheduling method provided in the above embodiments, in one possible implementation, task scheduling can also be implemented in the following manner. For example, taking the reimbursement document as the target task, the process can be described as follows: Step 1: Obtain the risk level of the expense reimbursement documents.

[0103] For example, this step can build a risk classification and identification model for multi-source unstructured text in expense reimbursement documents, then comprehensively analyze the structured fields and unstructured attachment text of the expense reimbursement documents, further extract hidden risk features and output accurate document risk levels, which serve as the preliminary judgment basis for adaptive switching of dispatch mode.

[0104] Step 2: Filter the initial object set according to the adjusted parameter set (i.e., the target matching parameter set) to obtain the first object set.

[0105] For example, this step can construct a four-dimensional dynamic weighted matching rule that integrates business scenarios (i.e., business scenario parameters), service objects (i.e., task object parameters), review positions (i.e., review authority parameters), and personnel professional profiles (i.e., professional competence parameters). The weight coefficients of each dimension are automatically and dynamically adjusted according to the risk level of the document to achieve accurate and differentiated matching between documents and reviewers.

[0106] As an example, for low-risk, ordinary expense reimbursement documents, the weight of the business scenario dimension is automatically increased, the logic for verifying job responsibilities and personnel professionalism is simplified, and the efficiency of batch dispatching large batches of regular documents is improved; for large-amount, abnormally high-risk documents, the weight of the three dimensions of service recipients, review positions, and personnel professional profiles is automatically increased, the matching of review responsibilities and verification of personnel professional capabilities are strengthened, and the financial risks caused by mismatch of high-risk documents are reduced from the source.

[0107] In one possible implementation, multi-dimensional historical work data of each auditor can be collected as core profiling factors (i.e., a set of historical profiling factors). These core profiling factors can include: historical document types (i.e., historical types), historical rejection error rates for various reimbursement documents (i.e., historical error rates), average processing time for large / risk documents (i.e., historical average processing time), and qualifications for handling specific reimbursement business such as travel / procurement / labor services (i.e., historical processing permissions). Furthermore, all core profiling factors can be weighted and calculated, and a specific business matching coefficient (i.e., professional competence coefficient) ranging from 0 to 100 can be output.

[0108] It should be understood that after determining the four-dimensional dynamic weighted matching rules, the initial object set can be filtered according to the four-dimensional dynamic weighted matching rules to obtain the first object set.

[0109] Step 3: Filter the first object set according to the load quota threshold to obtain the second object set.

[0110] For example, this step can design a three-tiered load management mechanism based on time-quantification, quantifying load quota values ​​according to document review time (i.e., review duration), risk cost (i.e., review risk), and human resource cost (i.e., review cost), configuring individual load thresholds and team total load thresholds, and providing a cross-position load adjustment pool to achieve real-time human resource load balancing management.

[0111] In one possible implementation, this step abandons the traditional document count mode and can quantify the load quota value of a single expense reimbursement document based on the review time, risk cost, and human resource cost of the expense reimbursement document. The load quota value can be divided into a basic load quota value and a risk load quota value. For example, a regular small-amount standardized expense reimbursement document occupies 1 basic load quota, while a large-amount / abnormal high-risk document occupies 2 risk load quotas.

[0112] Meanwhile, a three-tiered load threshold is set for each individual, with clear implementation rules: the basic load threshold is the baseline for batch assignment of regular documents, and personnel can normally accept regular documents if they do not reach this threshold; the risk load threshold is the upper limit for accepting high-risk documents, and once this threshold is reached, the acceptance of high-risk documents is restricted; the extreme warning threshold is the maximum load limit for a single person's work, and once this threshold is reached, all new work orders will be stopped.

[0113] Additionally, a team total load threshold and a cross-position load adjustment pool can be added to establish a two-layer load management logic: when a single person reaches the limit threshold, only the individual stops accepting orders; when the overall quota of a single position reaches the team's full load threshold, the order grabbing permission is automatically opened to related positions with cross-approval qualifications, realizing dynamic manpower replenishment between positions.

[0114] It should be understood that this step can integrate multi-dimensional data such as document business scenario, service object attributes, job responsibilities, personnel special matching coefficient, and personnel current cumulative load quota to complete the initial matching of work orders and reviewers, and simultaneously perform hierarchical load quota verification to filter out reviewers who exceed the load threshold, thus completing the pre-screening.

[0115] Step 4: Based on the preset task allocation strategy, determine the target object from the second object set and allocate the expense reimbursement documents to the target object.

[0116] For example, this step can integrate three scheduling modes: automatic order dispatch (i.e., automatic allocation mode), order grabbing dispatch (i.e., preemptive allocation mode), and manual order dispatch (i.e., manual allocation mode). Furthermore, it can adaptively switch the corresponding mode based on the risk level of the expense report, business type, job load status, or personnel cross-verification qualifications.

[0117] As an example, the automatic dispatch mode can be adapted to low-risk, standardized, routine documents. For instance, before a document enters the dispatch process, a multi-source unstructured text risk identification model is activated first to complete full-dimensional risk detection, including: automatically reading all attached text such as structured fields of reimbursement forms, invoice images and text, related business contracts, and reimbursement instructions, extracting hidden risk features such as excessive amounts, third-party cooperation expenditures, temporary special approvals, and abnormal consumption, and accurately determining the risk level of the document; furthermore, if it is determined to be a low-risk, standardized, routine reimbursement document, the automatic dispatch mode is automatically triggered.

[0118] At this point, the four-dimensional dynamic weighted matching rule automatically increases the weight of the business scenario dimension, simplifies redundant verification logic, and combines the personnel-specific matching coefficient, the current cumulative load quota and the tiered threshold to automatically allocate work orders according to the priority of the remaining available load of personnel from high to low. No manual intervention is required throughout the process, realizing the automated and balanced allocation of large batches of regular documents and efficiently adapting to standardized batch review scenarios.

[0119] As an example, the order-grabbing and dispatching mode can be adapted to medium-risk, cross-functional, general-purpose documents. For instance, a general-purpose document that is determined to be of medium risk through pre-assessment risk classification and can be cross-approved by multiple positions will trigger the order-grabbing and dispatching mode. The document can then be pushed to a shared task pool of personnel with the corresponding job permissions and cross-approval qualifications, allowing only personnel whose current cumulative load quota has not reached the basic load threshold to participate in order grabbing, thus ensuring a balanced workload.

[0120] It should be understood that, relying on the cross-position load adjustment pool mechanism, when the original position's overall quota is full and the team's load threshold is reached, the scope of order-grabbing personnel can be automatically expanded to related cross-approval positions. By having the approvers accept orders independently, manpower can be dynamically replenished, smoothing the overall load of each position and solving the problem of backlog of work orders.

[0121] As an example, the manual order dispatch mode can be adapted to high-risk, large-value, and complex orders. For instance, if an order is determined to be high-risk, large-value, abnormal, or complex after preliminary risk assessment, the manual order dispatch mode is locked, and the automatic dispatch and order-grabbing channels are turned off. Only management personnel can manually designate review positions and specific reviewers to complete targeted dispatch.

[0122] It should be understood that manual task assignment is not an arbitrary, subjective process. The system's task assignment interface can visualize in real time the specialized matching coefficients of all reviewers, as well as the current cumulative load quota or load threshold occupancy status, providing quantitative data references for manual assignment. Managers can prioritize personnel with high professional matching and low load to handle complex and high-risk documents, effectively avoiding problems such as timeouts in high-risk document review, insufficient professionalism, and concentrated task backlog. This is suitable for high-precision management scenarios such as specialized financial review and in-depth review of complex documents.

[0123] It should be understood that this step can adaptively switch between three scheduling modes—automatic dispatch, order-grabbing dispatch, and manual dispatch—based on parameters such as the prior risk level of the expense report, the business scenario, the attributes of the service recipient, the job load status, and the cross-qualification of personnel. The three modes have clear scenario boundaries and no overlap: low-risk routine documents are matched with automatic dispatch, medium-risk general documents are matched with order-grabbing dispatch, and high-risk difficult documents are matched with manual dispatch.

[0124] All dispatch processes and every personnel matching action are simultaneously subject to the rigid constraints of four-dimensional dynamic weighted matching rules and work hour quantification layered load quotas, forming a fixed and unified closed-loop intelligent dispatch scheduling system: document pre-risk identification - four-dimensional dynamic weighted matching - work hour layered load verification - multi-mode adaptive switching - cross-position manpower adjustment - work order closed-loop dispatch, fully covering all expense reimbursement review scenarios, achieving accurate work order dispatch, true manpower load balance, and flexible scheduling across all scenarios.

[0125] It should be understood that the task scheduling method provided in the embodiments of this application includes at least one of the following technical advantages: First, by constructing a four-dimensional dynamic weighted matching framework combined with a quantitative personnel profiling system, the limitations of traditional fixed matching are overcome, significantly improving the accuracy of dispatching. Traditional dispatching relies solely on the permissions of a single position to complete a coarse match with fixed and rigid weights. This invention adds a fourth matching dimension—personnel professional profiling—to construct a four-dimensional dynamic weighted matching system, which can dynamically adjust the weight and priority of each dimension in real time according to the risk level of the document. Simultaneously, by quantifying historical work data of personnel, a 0-100 score special business matching coefficient is generated, enabling refined and differentiated matching of the professional capabilities of personnel in the same position. This completely solves the core pain point of traditional technology—"mismatching high-risk documents with ordinary auditors and wasting high-quality human resources"—improving the professionalism of financial auditing from the underlying technical level and avoiding the risks of human error and oversight.

[0126] Secondly, a tiered workload management system based on quantified work hours, coupled with a cross-functional allocation mechanism, addresses industry-wide challenges such as distorted workload statistics and manpower imbalance. Abandoning the industry-standard, simplistic document counting-based workload statistics model, this system innovatively quantifies workload quotas based on document review hours and risk verification costs. It sets three-tiered workload thresholds for individuals and overall team workload thresholds, accurately differentiating the manpower costs associated with routine and high-risk documents. Simultaneously, a cross-functional workload allocation pool is established, providing triple protection through individual workload management, team workload management, and cross-functional manpower allocation. This completely resolves the issue of distorted workload caused by "piles of small-value documents and backlogs of high-risk documents," achieving dynamic and balanced scheduling of review manpower across the entire domain.

[0127] Third, a pre-emptive unstructured risk classification and identification system is added to achieve risk-based order control and proactively strengthen the financial compliance defense line. Breaking away from the traditional, simplistic risk identification model that relies solely on structured fields to classify documents, a multi-source unstructured text risk identification module is added at the very beginning of the order dispatch process. This module automatically parses implicit risk information from invoices, contracts, expense reimbursement notes, etc., accurately extracts hidden risk characteristics, and classifies documents into three risk levels. Based on the risk level, the corresponding order dispatch mode is precisely locked, preventing high-risk, abnormal, and complex documents from entering the automatic batch order dispatch process. This proactively manages financial risks at the source of order dispatch, significantly reducing corporate financial compliance risks.

[0128] Fourth, it achieves deep integration of three-mode adaptive scheduling with a full-dimensional technology system, fully covering financial audit scenarios. Addressing the shortcomings of traditional dispatching models, such as rigidity and poor scenario adaptability, it integrates three differentiated dispatching modes: automatic, order-grabbing, and manual, achieving fully automatic adaptive switching based on document risk levels. Core technologies such as four-dimensional weighted matching, quantified personnel profiling, hierarchical load management, and pre-emptive risk identification are fully embedded into every dispatching process, comprehensively covering all financial operation scenarios, including standardized batch auditing, cross-functional manpower supplementation, high-risk special review, and full-load adjustment of positions. This completely solves the industry pain points of traditional work order backlog, rigid scheduling, and incomplete scenario adaptability.

[0129] Fifth, it balances intelligent efficiency improvement with refined compliance management, highly meeting the core financial needs of enterprises. This invention implements differentiated scheduling logic for documents with different risk levels, achieving a two-way balance between efficiency and compliance: low-risk routine documents rely on a fully automated dispatch mode to reduce manual intervention and achieve efficient flow of large batches of work orders, reducing costs and improving efficiency; medium-to-high-risk, complex, and large-amount documents are matched with highly professional reviewers through a bidding and manual targeted dispatch mode, achieving refined and professional review. While realizing the intelligent upgrade of financial auditing, it fully meets the core needs of enterprises for compliance management and risk control of expense reimbursement, with strong technical applicability and industry adaptability.

[0130] Based on the task scheduling method provided in the above embodiments, see [link to relevant documentation]. Figure 2This application also provides a schematic diagram of the structure of a task scheduling device.

[0131] Combination Figure 2 As shown, the task scheduling device 20 provided in this application embodiment includes: Acquisition unit 21 is used to acquire the target risk level corresponding to the target task to be assigned; The adjustment unit 22 is used to adjust the initial matching parameter set according to the target risk level to obtain the target matching parameter set. The initial matching parameter set includes at least one of the following initial matching parameters: business scenario parameters, task object parameters, review authority parameters, or professional capability parameters. The first filtering unit 23 is used to filter the initial object set according to the target matching parameter set to obtain the first object set. The initial object set includes multiple initial objects, and the first object set includes at least one first object, which is an initial object that matches the target matching parameter set. The second filtering unit 24 is used to filter the first object set according to the load quota threshold to obtain a second object set. The second object set includes at least one second object, and the second object is the first object that meets the load quota threshold. The load quota threshold represents the upper limit of the load quota corresponding to the first object. The task scheduling unit 25 is used to assign a target task to a target object based on a preset task allocation strategy. The target object is a second object in a second object set.

[0132] In one possible implementation, the acquisition unit 21 is used for: Obtain the target task information corresponding to the target task; Feature extraction processing is performed on the target task information to obtain target risk features; The target risk level is determined based on the target risk characteristics and the preset risk level rules.

[0133] In one possible implementation, the adjustment unit 22 is used for: Determine the target weight coefficients for each initial matching parameter in the initial matching parameter set based on the target risk level; The target matching parameter set is determined based on each initial matching parameter and its corresponding target weight coefficient.

[0134] In one possible implementation, the second filtering unit 24 is used for: Obtain the basic load quota value corresponding to each first object. The basic load quota value includes the individual load quota value and the team load quota value. The individual load quota value is determined based on at least one of the audit duration, audit risk, or audit cost. The team load quota value represents the sum of the basic load quota values ​​of all corresponding objects. Select the first object from the first object set that meets the negative basic load quota condition to obtain the second object set; the basic load quota condition is that the individual load quota value is less than the first basic load quota value, and the team load quota value is less than the second basic load quota value.

[0135] In one possible implementation, the second filtering unit 24 is used for: Obtain the risk load quota value corresponding to each first object. The risk load quota value is determined based on at least one of the audit limit or audit risk. Select the first object from the first object set that meets the risk load quota condition to obtain the second object set; the risk load quota condition is that the risk load quota value is less than the risk load quota value.

[0136] In one possible implementation, the task scheduling unit 25 is used for: If the preset task allocation strategy is automatic allocation mode, the target task will be assigned to the highest priority second object in the second object set; If the preset task allocation strategy is preemptive allocation mode, then the target task will be allocated to the second object in the second object set that triggered the preemption operation; If the preset task allocation strategy is manual allocation mode, the target task will be allocated to the second object in the second object set that triggered the specified operation.

[0137] In one possible implementation, it also includes: a parameter determination unit, used to determine the professional competence parameters corresponding to the initial object in the following way: Obtain the set of historical profile factors corresponding to the initial object. The set of historical profile factors includes at least one of the following: historical task type, historical error rate, historical average processing time, or historical processing permissions. By performing a weighted calculation on all historical portrait factors in the historical portrait factor set, the professional ability parameters corresponding to the initial object are obtained.

[0138] It should be noted that the task scheduling device provided in this application embodiment has the same beneficial effects as the task scheduling method provided in the above embodiments, and therefore will not be described again.

[0139] In one possible implementation, see Figure 3 The figure is a schematic diagram of a control device provided in an embodiment of this application.

[0140] The control device may include a memory 311 and a processor 312. For example... Figure 3 As 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.

[0141] 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 task scheduling methods. The memory 311 can also store data.

[0142] 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)).

[0143] 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.

[0144] 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.

[0145] 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.

[0146] 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 task scheduling method, characterized in that, The method includes: Obtain the target risk level corresponding to the target task to be assigned; The initial matching parameter set is adjusted according to the target risk level to obtain the target matching parameter set. The initial matching parameter set includes at least one of the following initial matching parameters: business scenario parameters, task object parameters, review permission parameters, or professional capability parameters. The initial object set is filtered according to the target matching parameter set to obtain a first object set. The initial object set includes multiple initial objects, and the first object set includes at least one first object, which is an initial object that conforms to the target matching parameter set. The first object set is filtered according to the load quota threshold to obtain a second object set. The second object set includes at least one second object, which is a first object that meets the load quota threshold. The load quota threshold represents the upper limit of the load quota corresponding to the first object. Based on a preset task allocation strategy, the target task is assigned to a target object, which is a second object in the second object set.

2. The task scheduling method according to claim 1, characterized in that, The process of obtaining the target risk level corresponding to the target task to be assigned includes: Obtain the target task information corresponding to the target task; The target task information is processed by feature extraction to obtain target risk features; The target risk level corresponding to the target task is determined based on the target risk characteristics and the preset risk level rules.

3. The task scheduling method according to claim 1, characterized in that, The step of adjusting the initial matching parameter set according to the target risk level to obtain the target matching parameter set includes: Determine the target weight coefficient corresponding to each initial matching parameter in the initial matching parameter set based on the target risk level; The target matching parameter set is determined based on each of the initial matching parameters and their respective target weight coefficients.

4. The task scheduling method according to claim 1, characterized in that, The step of filtering the first object set according to the load quota threshold to obtain the second object set includes: Obtain the basic load quota value corresponding to each of the first objects. The basic load quota value includes an individual load quota value and a team load quota value. The individual load quota value is determined based on at least one of the following: review duration, review risk, or review cost. The team load quota value represents the sum of the basic load quota values ​​of all corresponding objects. The first object set that meets the negative basic load quota condition is selected from the first object set to obtain the second object set; the basic load quota condition is that the individual load quota value is less than the first basic load quota value, and the team load quota value is less than the second basic load quota value.

5. The task scheduling method according to claim 1, characterized in that, The step of filtering the first object set according to the load quota threshold to obtain the second object set includes: Obtain the risk load quota value corresponding to each of the first objects, wherein the risk load quota value is determined based on at least one of audit limit or audit risk; The first object set that meets the risk load quota condition is selected from the first object set to obtain the second object set; the risk load quota condition is that the risk load quota value is less than the risk load quota value.

6. The task scheduling method according to claim 1, characterized in that, The process of assigning the target task to the target object based on a preset task allocation strategy includes: If the preset task allocation strategy is automatic allocation mode, then the target task is allocated to the second object with the highest priority in the second object set; If the preset task allocation strategy is a preemptive allocation mode, then the target task is allocated to the second object in the second object set that triggered the preemptive operation; If the preset task allocation strategy is manual allocation mode, then the target task is allocated to the second object in the second object set that triggers the specified operation.

7. The task scheduling method according to any one of claims 1-6, characterized in that, This also includes determining the professional competence parameters corresponding to the initial object in the following ways: Obtain the set of historical profile factors corresponding to the initial object. The set of historical profile factors includes at least one of the following historical profile factors: historical task type, historical error rate, historical average processing time, or historical processing permissions. The professional competence parameters corresponding to the initial object are obtained by weighting all historical portrait factors in the set of historical portrait factors.

8. A task scheduling device, characterized in that, The device includes: The acquisition unit is used to acquire the target risk level corresponding to the target task to be assigned. The adjustment unit is used to adjust the initial matching parameter set according to the target risk level to obtain the target matching parameter set. The initial matching parameter set includes at least one of the following initial matching parameters: business scenario parameters, task object parameters, review authority parameters, or professional capability parameters. The first filtering unit is used to filter the initial object set according to the target matching parameter set to obtain a first object set. The initial object set includes multiple initial objects, and the first object set includes at least one first object, which is an initial object that conforms to the target matching parameter set. The second filtering unit is used to filter the first object set according to the load quota threshold to obtain a second object set. The second object set includes at least one second object, and the second object is a first object that meets the load quota threshold. The load quota threshold represents the upper limit of the load quota corresponding to the first object. The task scheduling unit is used to assign the target task to a target object based on a preset task allocation strategy. The target object is a second object in the second object set.

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 perform the task scheduling method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The system contains a computer program that is loaded by a processor to execute the task scheduling method as described in any one of claims 1-7.

Images