Complex business project auditing system based on multi-dimensional historical data

Abstract

The invention discloses a complex business project auditing system based on multi-dimensional historical data, and belongs to the technical field of business project auditing, and the system comprises a data collection and preprocessing module which is used for collecting business project auditing multi-source data and carrying out the preprocessing of the business project auditing multi-source data; the risk assessment anomaly detection module is used for identifying an abnormal mode and a potential risk in the auditing task and determining a business project auditing risk assessment result; the task sorting and intelligent distribution module is used for carrying out priority sorting on the tasks and intelligently distributing the tasks; and the real-time monitoring dynamic adjustment module is used for dynamically adjusting the task allocation scheduling scheme according to the real-time monitoring data feedback. According to the method and the device, the problem of low auditing efficiency caused by incapability of effectively identifying abnormal modes and potential risks in auditing tasks and incapability of optimizing resource allocation in the prior art is solved. According to the method, the abnormal mode and the potential risk in the auditing task can be effectively identified, resource allocation can be optimized, the risk can be reduced, and the auditing efficiency of complex business items can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of business project review, and in particular to a complex business project review system based on multi-dimensional historical data. Background Art

[0002] Business project review refers to a comprehensive review of project activities based on relevant laws and regulations, corporate policies and management standards to determine whether they meet the expected goals, whether there are potential problems, and to make improvement suggestions.

[0003] The importance of business project review is reflected in the following aspects: 1) Ensuring compliance: ensuring that the project complies with national laws and regulations and corporate policies; 2) Optimizing resource allocation: rationally allocating resources to avoid waste; 3) Improving project efficiency: improving project execution efficiency by discovering and correcting problems; 4) Supporting decision-making: providing management with a scientific basis for decision-making.

[0004] The existing complex business project audits cannot effectively identify abnormal patterns and potential risks in the audit tasks, cannot optimize resource allocation and reduce risks, resulting in low efficiency in complex business project audits. Summary of the Invention

[0005] The purpose of the present invention is to provide a complex business project audit system based on multi-dimensional historical data, which can effectively identify abnormal patterns and potential risks in audit tasks, optimize resource allocation and reduce risks, improve the efficiency of complex business project audits, and solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: A complex business project review system based on multi-dimensional historical data, including: The data collection and preprocessing module is used to collect multi-source data of business project review from the enterprise's internal system and external data sources, and pre-process the multi-source data of business project review to determine the characteristic data of business project review; The risk assessment anomaly detection module is used to build a business project review risk assessment model using machine learning algorithms, identify abnormal patterns and potential risks in review tasks, and determine the business project review risk assessment results; Task sorting and intelligent allocation module, used to prioritize tasks and intelligently allocate tasks to ensure efficient use of resources; Real-time monitoring and dynamic adjustment module, used to monitor the task execution status in real time and dynamically adjust the task allocation and scheduling plan based on real-time monitoring data feedback to respond to emergencies or resource changes; The data visualization decision support module is used to use data visualization tools to display audit progress, risk distribution and resource usage to facilitate management decision-making.

[0007] Preferably, to identify abnormal patterns and potential risks in the audit tasks and determine the business project audit risk assessment results, the following operations are performed: Deploy the business project review risk assessment model and deploy the business project review risk assessment model in the actual business project review risk assessment environment; Input the business project audit feature data into the business project audit risk assessment model, analyze the business project audit feature data according to the business project audit risk assessment model, and automatically identify abnormal patterns and potential risks in business project audit tasks, assess the task risk level, and determine the business project audit risk assessment results.

[0008] Preferably, to assess the risk level of a task, perform the following operations: Extracting the number of associated business nodes corresponding to the abnormal pattern obtained by analyzing the business project review risk assessment model and the abnormality occurrence probability corresponding to the associated business nodes; Retrieve the weight value corresponding to each associated business node; The weight value corresponding to each associated business node and the probability of abnormal occurrence corresponding to the associated business node are used for weighted average processing to obtain the comprehensive parameter of the associated business abnormal probability. ; Where a represents the number of associated business nodes; w yi represents the weight value corresponding to the i-th associated business node; P yi represents the probability of anomaly occurrence corresponding to the i-th associated business node; Extract the number of business process links involved in the potential risks obtained through analysis of the business project review risk assessment model and the corresponding risk occurrence probability of each business process link; Retrieve the weight value corresponding to each business process link; The weight value corresponding to each business process link and the risk probability corresponding to each business process link are used for weighted average processing to obtain the comprehensive parameter of business process risk probability. ; Where b represents the number of business process links; w i Indicates the weight value corresponding to the i-th business process link; P xi represents the probability of risk occurrence corresponding to the i-th business process link; The associated business anomaly probability comprehensive parameter and the business process risk probability comprehensive parameter are summed to obtain an initial risk score corresponding to the business project review feature data; Compare the initial risk score corresponding to the business project review feature data with the preset risk score threshold; When the initial risk score corresponding to the business project review feature data does not exceed the preset risk score threshold, the business project review task is determined to be at a low risk level; When the initial risk score corresponding to the business project review feature data exceeds a preset risk score threshold, the business project review feature data is subjected to a secondary risk level classification.

[0009] Preferably, when the initial risk score corresponding to the business project review feature data exceeds a preset risk score threshold, the business project review feature data is subjected to a secondary risk level classification, and the following operations are performed: Retrieve the number of associated business nodes and the probability of abnormal occurrence corresponding to the associated business nodes; Obtaining a standard deviation of node anomaly probability based on the number of associated service nodes and the anomaly occurrence probability corresponding to the associated service nodes; Retrieve the risk probability corresponding to each business process link; The risk probability of each business process link corresponding to the risk probability threshold is compared with the preset risk probability threshold, wherein the preset risk probability threshold value is 0.4; Screening out business process links whose risk occurrence probability is not lower than the preset risk probability threshold as target business process links; Determine whether non-target business process links are interspersed between target business process links, and obtain each group of target business process links interspersed with non-target business process links; Set the risk impact coefficient based on each group of target business process links that are mixed with non-target business process links; The risk impact coefficient is combined with the node abnormality probability standard deviation to obtain the secondary risk determination parameter; Comparing the secondary risk determination parameter with a preset risk determination parameter threshold; When the secondary risk determination parameter is lower than the preset risk determination parameter threshold, the business project review task is determined to be of medium risk level; When the secondary risk determination parameter is not lower than a preset risk determination parameter threshold, the business project review task is determined to be at a high risk level.

[0010] Preferably, tasks are prioritized and intelligently assigned, performing the following operations: Prioritize complex business project review tasks based on the business project review risk assessment results, combining risk level factors, task complexity factors, and urgency factors to determine task priority; Dynamically assign tasks based on task complexity, reviewer capabilities, work status, and task priority to optimize resource allocation.

[0011] Preferably, the multi-source data of the business project review is pre-processed by performing the following operations: Clean the multi-source data of business project audits, remove noise irrelevant to complex business project audits, and process missing values ​​and outliers in the multi-source data of business project audits; The multi-source data of business project review is converted into a unified data format, the dimensional differences in the multi-source data of business project review are removed, and standardized multi-source data of business project review is formed.

[0012] Preferably, the multi-source data of the business project review is pre-processed, and the following operations are further performed: Integrate multi-source data of business project review, merge multi-source data of business project review from different sources into a unified data view, and store the integrated multi-source data of business project review; Perform feature extraction on multi-source data of business project review, extract features related to complex business project review from the multi-source data of business project review, and determine business project review feature data.

[0013] Preferably, a business project review risk assessment model is constructed using a machine learning algorithm to perform the following operations: Collecting historical data on business project reviews and dividing the historical data into a training set and a test set; Based on machine learning algorithms, a training set is used to train the machine learning model, allowing the machine learning model to autonomously learn complex business project audit behaviors from the training set, identify abnormal patterns and potential risks in audit tasks, and determine a business project audit risk assessment model based on machine learning; Input the test set into the business project review risk assessment model, test the business project review risk assessment model based on the test set, evaluate the performance of the business project review risk assessment model, and determine the model test evaluation results; The business project review risk assessment model is adjusted and optimized based on the model test evaluation results to determine the optimal business project review risk assessment model.

[0014] Preferably, to evaluate the performance of the business project review risk assessment model, the following operations are performed: Use the test set to test the business project audit risk assessment model, and judge whether the business project audit risk assessment model can achieve the expected effect of identifying abnormal patterns and potential risks in audit tasks based on the accuracy, precision and recall rate; When the business project audit risk assessment model cannot achieve the expected effect of identifying abnormal patterns and potential risks in the audit task, the parameters of the business project audit risk assessment model are adjusted and optimized until the business project audit risk assessment model can achieve the expected effect of identifying abnormal patterns and potential risks in the audit task, thereby determining the optimal business project audit risk assessment model.

[0015] Preferably, use data visualization tools to display audit progress, risk distribution, and resource usage, and perform the following operations: Use data visualization tools to intuitively display audit progress, risk distribution, and resource usage through a variety of charts and interactive dashboards; Regularly generate complex business project audit reports based on audit progress, risk distribution, and resource usage to provide management with detailed information on task allocation and resource utilization, helping managers quickly identify risk hotspots, optimize scheduling plans, and provide data support for strategic decision-making.

[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention collects multi-source data on business project audits from the enterprise's internal systems and external data sources, pre-processes the multi-source data on business project audits, determines business project audit feature data, and uses machine learning algorithms to build a business project audit risk assessment model to identify abnormal patterns and potential risks in audit tasks and determine business project audit risk assessment results. According to the business project audit risk assessment results and in combination with risk level factors, task complexity factors and urgency factors, complex business project audit tasks are prioritized to determine task priorities, and tasks are dynamically allocated according to task complexity, auditor capabilities and work status and in combination with task priorities, thereby optimizing resource allocation and ensuring efficient use of resources. This can effectively identify abnormal patterns and potential risks in audit tasks, optimize resource allocation and reduce risks, and improve the efficiency of complex business project audits.

[0017] 2. The present invention monitors the task execution status in real time and dynamically adjusts the task allocation and scheduling plan based on the real-time monitoring data feedback to respond to emergencies or resource changes. It also uses data visualization tools to intuitively display the audit progress, risk distribution and resource usage through a variety of charts and interactive dashboards. It regularly generates complex business project audit reports based on the audit progress, risk distribution and resource usage to provide management with detailed information on task allocation and resource utilization, help managers quickly identify risk hotspots, optimize scheduling plans, and provide data support for strategic decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1This is a module diagram of the complex business project review system based on multi-dimensional historical data of the present invention. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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 are within the scope of protection of the present invention.

[0020] To address the inefficiency of complex project audits due to the inability to effectively identify abnormal patterns and potential risks in audit tasks, optimize resource allocation, and reduce risks, please refer to Figure 1 , this embodiment provides the following technical solutions: A complex business project review system based on multi-dimensional historical data, including: The data collection and preprocessing module is used to collect multi-source data of business project review from the enterprise's internal system and external data sources, and pre-process the multi-source data of business project review to determine the business project review feature data.

[0021] It should be noted that internal enterprise systems include ERP, CRM, etc., and external data sources include industry databases, social media, etc.

[0022] In this embodiment, the multi-source data of business project review is pre-processed by performing the following operations: Clean the multi-source data of business project review, remove the noise irrelevant to the complex business project review, and process the missing values ​​and outliers in the multi-source data of business project review, which can improve the data quality of the multi-source data of business project review; Convert the multi-source data of business project review into a unified data format, remove the dimensional differences in the multi-source data of business project review, and form standardized multi-source data of business project review to improve data availability and facilitate subsequent data analysis.

[0023] In this embodiment, the multi-source data of business project review is pre-processed, and the following operations are also performed: Integrate multi-source data of business project review, merge multi-source data of business project review from different sources into a unified data view, and store the integrated multi-source data of business project review; Perform feature extraction on multi-source data of business project audits, extract features related to complex business project audits from multi-source data of business project audits, determine business project audit feature data, and facilitate subsequent analysis of business project audit feature data to identify abnormal patterns and potential risks in audit tasks.

[0024] The risk assessment anomaly detection module is used to build a business project review risk assessment model using machine learning algorithms, identify abnormal patterns and potential risks in review tasks, and determine the business project review risk assessment results.

[0025] In this embodiment, a business project review risk assessment model is constructed using a machine learning algorithm to perform the following operations: Collecting historical data on business project reviews and dividing the historical data into a training set and a test set; Based on machine learning algorithms, a training set is used to train the machine learning model, allowing the machine learning model to autonomously learn complex business project audit behaviors from the training set, identify abnormal patterns and potential risks in audit tasks, and determine a business project audit risk assessment model based on machine learning; Input the test set into the business project review risk assessment model, test the business project review risk assessment model based on the test set, evaluate the performance of the business project review risk assessment model, and determine the model test evaluation results; Among them, the test set is used to test the business project audit risk assessment model, and the accuracy, precision and recall rate are used to judge whether the business project audit risk assessment model can achieve the expected effect of identifying abnormal patterns and potential risks in the audit task; When the business project audit risk assessment model cannot achieve the expected effect of identifying abnormal patterns and potential risks in the audit task, the parameters of the business project audit risk assessment model are adjusted and optimized until the business project audit risk assessment model can achieve the expected effect of identifying abnormal patterns and potential risks in the audit task, thereby determining the optimal business project audit risk assessment model.

[0026] In this embodiment, to identify abnormal patterns and potential risks in audit tasks and determine the business project audit risk assessment results, the following operations are performed: Deploy the business project review risk assessment model and deploy the business project review risk assessment model in the actual business project review risk assessment environment; Input the business project audit feature data into the business project audit risk assessment model, analyze the business project audit feature data according to the business project audit risk assessment model, and automatically identify abnormal patterns and potential risks in business project audit tasks, assess the task risk level, and determine the business project audit risk assessment results.

[0027] The task sorting and intelligent allocation module is used to prioritize tasks and intelligently allocate tasks to ensure efficient use of resources.

[0028] Specifically, to assess the risk level of a task, perform the following operations: Extracting the number of associated business nodes corresponding to the abnormal pattern obtained by analyzing the business project review risk assessment model and the abnormality occurrence probability corresponding to the associated business nodes; Retrieve the weight value corresponding to each associated business node; The weight value corresponding to each associated business node and the probability of abnormal occurrence corresponding to the associated business node are used for weighted average processing to obtain the comprehensive parameter of the associated business abnormal probability. ; Where a represents the number of associated business nodes; w yi represents the weight value corresponding to the i-th associated business node. At the same time, the corresponding ratio of the weight value is set based on experience in combination with the importance of the associated business node, for example, 0.1 or 0.18, and the sum of the weight values ​​corresponding to all associated business nodes is 1; yi represents the probability of anomaly occurrence corresponding to the i-th associated business node; Extract the number of business process links involved in the potential risks obtained through analysis of the business project review risk assessment model and the corresponding risk occurrence probability of each business process link; Retrieve the weight value corresponding to each business process link; The weight value corresponding to each business process link and the risk probability corresponding to each business process link are used for weighted average processing to obtain the comprehensive parameter of business process risk probability. ; Where b represents the number of business process links; w i represents the weight value corresponding to the i-th business process link. At the same time, the corresponding ratio of the weight value is set based on experience in combination with the importance of the business process link, for example, 0.2 or 0.21, and the sum of the weight values ​​corresponding to all related business nodes is 1; P xi represents the probability of risk occurrence corresponding to the i-th business process link; The associated business anomaly probability comprehensive parameter and the business process risk probability comprehensive parameter are summed to obtain an initial risk score corresponding to the business project review feature data; Compare the initial risk score corresponding to the business project review feature data with the preset risk score threshold; When the initial risk score corresponding to the business project review feature data does not exceed the preset risk score threshold, the business project review task is determined to be at a low risk level; When the initial risk score corresponding to the business project review feature data exceeds a preset risk score threshold, the business project review feature data is subjected to a secondary risk level classification.

[0029] The technical solution achieves the following: By extracting the number of associated business nodes and anomaly probability corresponding to anomaly patterns, the number of business process links involved in potential risks and their probability of occurrence, and then performing quantitative calculations based on the weights of the nodes and links, abstract anomalies and risks are transformed into measurable comprehensive parameters of associated business anomaly probability, comprehensive parameters of business process risk probability, and an initial risk score. This allows for the precise capture and quantification of hidden and highly correlated anomaly patterns and potential risks in complex business projects, avoiding the subjectivity and omissions inherent in manual identification. Furthermore, based on the initial risk score and the partitioning logic based on the preset threshold, low-risk tasks can be allocated basic audit resources for rapid processing, while tasks exceeding the threshold are placed in a secondary partitioning process to match more precise resource allocations. This avoids redundant resource consumption on low-risk tasks while ensuring that high-risk tasks receive sufficient audit resources, achieving a reasonable allocation of resources to high-risk, high-priority audit tasks. Furthermore, by defining risk as a definitive score through quantitative assessment, the potential risks of low-risk tasks can be systematically monitored, while high-risk tasks exceeding the threshold are prioritized through secondary partitioning. This reduces audit oversights caused by untimely risk identification or inadequate risk management, thereby reducing the likelihood of risks becoming actual problems. At the same time, quantitative parameters and scoring mechanisms are used to achieve an orderly division of risk levels, avoiding the inefficiency caused by the use of a unified review process for all tasks. Low-risk tasks can be passed quickly and high-risk tasks can be accurately focused on, shortening the overall review cycle. At the same time, repeated reviews and invalid operations are reduced, and the operating efficiency of the review process in complex business scenarios is improved.

[0030] Specifically, when the initial risk score corresponding to the business project review feature data exceeds a preset risk score threshold, the business project review feature data is subjected to a secondary risk level classification, and the following operations are performed: Retrieve the number of associated business nodes and the probability of abnormal occurrence corresponding to the associated business nodes; Obtaining a standard deviation of node anomaly probability based on the number of associated service nodes and the anomaly occurrence probability corresponding to the associated service nodes; Retrieve the risk probability corresponding to each business process link; The risk probability of each business process link corresponding to the risk probability threshold is compared with the preset risk probability threshold, wherein the preset risk probability threshold value is 0.4; Screening out business process links whose risk occurrence probability is not lower than the preset risk probability threshold as target business process links; Determine whether non-target business process links are interspersed between target business process links, and obtain each group of target business process links interspersed with non-target business process links; Set the risk impact coefficient based on each group of target business process links that are mixed with non-target business process links; The risk impact coefficient is obtained by the following formula:

[0031] Where S represents the risk impact coefficient; n represents the number of target business process links with non-target business process links between the process links; w 01i and w 02i They represent the weight values ​​of the first target business process link and the second target business process link included in the i-th group of target business process links; P 01i and P 02i They represent the risk probabilities of the first target business process link and the second target business process link contained in the i-th group of target business process links; P fi Represents the weighted average probability value corresponding to the risk probabilities and weight values ​​of all non-target business process links interspersed between the target business process links in the i-th group; The risk impact coefficient is combined with the node abnormality probability standard deviation to obtain the secondary risk determination parameter; The secondary risk determination parameter is obtained by the following formula:

[0032] Among them, F represents the secondary risk judgment parameter; S represents the risk impact coefficient; σ represents the standard deviation of the node abnormality probability; P ymax and P ymin Respectively represent the maximum and minimum values ​​of the anomaly occurrence probability in the associated business node; Comparing the secondary risk determination parameter with a preset risk determination parameter threshold; When the secondary risk determination parameter is lower than the preset risk determination parameter threshold, the business project review task is determined to be of medium risk level; When the secondary risk determination parameter is not lower than a preset risk determination parameter threshold, the business project review task is determined to be at a high risk level.

[0033] The technical solution described above achieves the following: Through detailed analysis of related business nodes and business process links during secondary segmentation (such as calculating the standard deviation of node anomaly probabilities and screening target business process links), it uncovers more subtle risk association patterns within business processes (such as the risk transmission characteristics of target links interspersed with non-target links). This addresses the initial assessment's inability to capture risks in complex processes, enabling more comprehensive and in-depth identification of anomaly patterns and potential risks, and enabling the system to more accurately understand complex business risks. For example, within a multi-linked business process, it can identify clusters and interval distribution characteristics of key risk links, uncovering risk transmission chains that are often overlooked by traditional single-link assessments. Secondary segmentation determines target links and risk impact coefficients based on risk probability and link distribution, providing a more granular basis for resource allocation. High-risk characteristics (such as target link groups interspersed with non-target links) are assigned higher resource priority, while low-risk characteristics are rationally streamlined. This allows system resource allocation to adapt to the complex business risk hierarchy and avoid resource misallocation. For example, for high-impact tasks interspersed with multiple non-target links, experienced reviewers are assigned and the review time is extended; for tasks with simpler risk characteristics, standardized and expedited processes are used. At the same time, a risk impact coefficient formula is introduced to quantify the risk amplification effect caused by the inclusion of process links. Combined with the secondary judgment parameters, the management and control of high-risk tasks can be more focused on key risk points (such as cluster risks in target links). The system can formulate special management and control strategies for these key risk points (such as multiple rounds of review and cross-departmental collaborative verification) to reduce the actual probability of risk occurrence and strengthen the ability to resist complex business risks. In addition, the secondary division avoids the problem of low audit accuracy caused by the rigid audit of complex businesses by accurately identifying high-risk characteristic tasks and matching them with adaptive processes. Low-risk tasks continue to use the efficient initial process, and high-risk tasks focus on the audit of key risk points, reducing invalid audit links, improving the overall flow efficiency of the system in handling complex businesses, and making audit resources more reasonably allocated to tasks of different risk levels, shortening the overall business audit cycle.

[0034] On the other hand, the initial risk score was derived through a weighted average of nodes and processes, focusing on the average impact of fundamental risk factors. The secondary categorization approach introduced new dimensions, such as the distribution characteristics of process links (the inclusion of target links) and the standard deviation of node anomaly probabilities, to supplement the dynamic characteristics of risk transmission and fluctuation within the process. From static average assessment to dynamic fluctuation and distribution characteristic assessment, this approach covers the diverse attributes of complex business risks, making risk level assessment more consistent with the overall picture of actual business risks. For example, during project audits, the initial score reflects the average risk level of each link. The secondary categorization identified the risk resonance effect caused by the clustered distribution of key risk links, correcting the bias of focusing solely on the average.

[0035] At the same time, secondary categorization refines the criteria for identifying high-risk tasks (for example, by comparing F with a threshold to distinguish between medium and high risk). Combined with the basic screening provided by initial scoring, this system establishes a risk level boundary based on "initial coarse screening + secondary fine screening." For complex tasks, this not only prevents initial scoring from misjudging high-risk tasks, but also allows for precise differentiation between medium and high risk through secondary categorization, making risk level boundaries clearer and more accurate, thereby reducing misjudgments. For example, in large-scale project bidding and tendering reviews, after initial scoring identifies high risk, secondary categorization precisely distinguishes between "medium risk" (F slightly exceeds the threshold, indicating manageable risk) and "high risk" (F far exceeds the threshold, indicating a high risk of uncontrollable risk), providing an accurate basis for subsequent handling. Furthermore, while the initial solution rapidly completes basic risk stratification (low / high risk initial screening), secondary categorization only conducts detailed assessments of high-risk tasks from the initial screening, avoiding complex secondary analysis for all tasks and reducing inefficient calculations. When the system handles complex tasks, the majority of low-risk tasks are quickly approved using the efficient initial process, leaving only a small number of high-risk tasks undergoing the time-consuming but necessary secondary evaluation, improving overall assessment efficiency.

[0036] Furthermore, the secondary partitioning formula focuses on key characteristics of high-risk tasks (such as the number of nodes and the complexity of links). Parameter acquisition relies on the results of previous model analysis (reusable initial assessment data such as the number of associated business nodes and the probability of risk occurrence), reducing the time and cost of repeated data collection. Furthermore, the formula's logical operation is clear and can be rapidly executed through algorithmic optimization (such as parallel calculation of node standard deviations and link probability comparisons), ensuring efficient assessment even under complex calculations. Furthermore, the secondary partitioning outputs clear medium- and high-risk level determinations (F comparison results against thresholds). Combined with the initial scoring, this creates a rapid mapping of "risk level to handling strategy." Based on these determinations, the system can directly trigger pre-set handling processes (such as review of medium-risk task initiation criteria and expert review of high-risk task initiation) without the need for additional decision-making. This accelerates the closed loop from risk assessment to control execution and improves the efficiency of handling complex business risks.

[0037] In this embodiment, tasks are prioritized and intelligently assigned by performing the following operations: Prioritize complex business project review tasks based on the business project review risk assessment results, combining risk level factors, task complexity factors, and urgency factors to determine task priority; Dynamically assign tasks based on task complexity, reviewer capabilities, work status, and task priority to optimize resource allocation.

[0038] It should be noted that task priorities are determined by combining risk level factors, task complexity factors, and urgency factors. Risk level factors refer to the degree of risk that a task may bring, such as financial loss, reputation damage, etc. Task complexity factors refer to the difficulty of the task and the professional skills required to complete it. Urgency factors refer to the urgency of the task to be completed, such as deadlines, customer needs, etc. Based on the risk level, task complexity, and urgency factors, a scoring scale (e.g., 1-5 points) is used to set scoring criteria for each key factor. For example: Risk level: 1 (low risk) - 5 (high risk); Task complexity: 1 (easy) - 5 (complex); Urgency: 1 (not urgent) - 5 (urgent); A risk matrix is ​​constructed based on the scoring criteria set for each key factor in the scoring scale. The risk matrix is ​​a two-dimensional table with risk level and urgency as rows and columns, and the intersection points represent the corresponding risk levels, as shown in Table 1:

[0039] In this matrix, numbers represent risk levels. The larger the number, the higher the risk. Therefore, task priorities are determined based on the business project review risk assessment results, combined with risk level factors, task complexity factors, and urgency factors. By prioritizing complex business project review tasks and assigning a priority to each task, tasks with high review risks can be reviewed first. At the same time, task priorities may change with project progress, resource changes, or the emergence of new information. Therefore, priorities need to be updated regularly or in real time to ensure that resource allocation and project progress match risk control.

[0040] It should be noted that tasks are dynamically allocated based on task complexity, auditor capabilities, work status, and task priority. Task complexity refers to the difficulty of the task and the professional skills required to complete it; auditor capabilities refer to the auditor's skills, including the auditor's expertise, experience level, and certification status in different fields; and work status refers to the auditor's workload, including the current number of tasks, task completion progress, and estimated completion time.

[0041] Dynamically match tasks and reviewers based on task priority and in combination with task complexity, reviewer capabilities and work status. At the same time, adjust task allocation in real time based on changes in work status and task priority.

[0042] Real-time monitoring and dynamic adjustment module, used to monitor the task execution status in real time and dynamically adjust the task allocation and scheduling plan based on real-time monitoring data feedback to respond to emergencies or resource changes; The data visualization decision support module is used to use data visualization tools to display audit progress, risk distribution and resource usage to facilitate management decision-making.

[0043] In this example, a data visualization tool is used to display the audit progress, risk distribution, and resource usage. The following operations are performed: Use data visualization tools to intuitively display audit progress, risk distribution, and resource usage through a variety of charts and interactive dashboards; Regularly generate complex business project audit reports based on audit progress, risk distribution, and resource usage to provide management with detailed information on task allocation and resource utilization, helping managers quickly identify risk hotspots, optimize scheduling plans, and provide data support for strategic decision-making.

[0044] In summary, a business project audit risk assessment model is constructed using machine learning algorithms to identify abnormal patterns and potential risks in audit tasks and determine the business project audit risk assessment results. Complex business project audit tasks are prioritized based on the business project audit risk assessment results and combined with risk level factors, task complexity factors, and urgency factors to determine task priority. Tasks are dynamically allocated based on task complexity, auditor capabilities, work status, and task priority, optimizing resource allocation and ensuring efficient resource utilization. This can effectively identify abnormal patterns and potential risks in audit tasks, optimize resource allocation, reduce risks, and improve the efficiency of complex business project audits.

[0045] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0046] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A complex business project review system based on multi-dimensional historical data, characterized by: include: The data collection and preprocessing module is used to collect multi-source data of business project review from the enterprise's internal system and external data sources, and pre-process the multi-source data of business project review to determine the characteristic data of business project review; The risk assessment anomaly detection module is used to build a business project review risk assessment model using machine learning algorithms, identify abnormal patterns and potential risks in review tasks, assess the risk level of tasks, and determine the business project review risk assessment results; The sum of the associated business anomaly probability comprehensive parameter and the business process risk probability comprehensive parameter is processed to obtain the initial risk score corresponding to the business project review feature data. When the initial risk score corresponding to the business project review feature data exceeds the preset risk score threshold, the business project review feature data is subjected to a secondary risk level classification. The risk impact coefficient is combined with the standard deviation of the node abnormality probability to obtain the secondary risk determination parameter, and the risk level of the business project review task is determined based on the secondary risk determination parameter; Task sorting and intelligent allocation module, used to prioritize tasks and intelligently allocate tasks to ensure efficient use of resources; Real-time monitoring and dynamic adjustment module, used to monitor the task execution status in real time and dynamically adjust the task allocation and scheduling plan based on real-time monitoring data feedback to respond to emergencies or resource changes; The data visualization decision support module is used to use data visualization tools to display audit progress, risk distribution and resource usage to facilitate management decision-making.

2. The complex business project review system based on multi-dimensional historical data according to claim 1 is characterized in that: Identify unusual patterns and potential risks in audit tasks and determine the business project audit risk assessment results by performing the following: Deploy the business project review risk assessment model and deploy the business project review risk assessment model in the actual business project review risk assessment environment; Input the business project audit feature data into the business project audit risk assessment model, analyze the business project audit feature data according to the business project audit risk assessment model, and automatically identify abnormal patterns and potential risks in business project audit tasks, assess the task risk level, and determine the business project audit risk assessment results.

3. The complex business project review system based on multi-dimensional historical data according to claim 2 is characterized in that: To assess the risk level of the task, do the following: Extracting the number of associated business nodes corresponding to the abnormal pattern obtained by analyzing the business project review risk assessment model and the abnormality occurrence probability corresponding to the associated business nodes; Retrieve the weight value corresponding to each associated business node; The weight value corresponding to each associated business node and the probability of abnormal occurrence corresponding to the associated business node are used for weighted average processing to obtain the comprehensive parameter of the associated business abnormality probability. ; Where a represents the number of associated business nodes; w yi represents the weight value corresponding to the i-th associated business node; P yi represents the probability of anomaly occurrence corresponding to the i-th associated business node; Extract the number of business process links involved in the potential risks obtained through analysis of the business project review risk assessment model and the corresponding risk occurrence probability of each business process link; Retrieve the weight value corresponding to each business process link; The weight value corresponding to each business process link and the risk probability corresponding to each business process link are used for weighted average processing to obtain the comprehensive parameter of business process risk probability. ; Where b represents the number of business process links; w i Indicates the weight value corresponding to the i-th business process link; P xi represents the probability of risk occurrence corresponding to the i-th business process link; The associated business anomaly probability comprehensive parameter and the business process risk probability comprehensive parameter are summed to obtain an initial risk score corresponding to the business project review feature data; Compare the initial risk score corresponding to the business project review feature data with the preset risk score threshold; When the initial risk score corresponding to the business project review feature data does not exceed the preset risk score threshold, the business project review task is determined to be at a low risk level; When the initial risk score corresponding to the business project review feature data exceeds a preset risk score threshold, the business project review feature data is subjected to a secondary risk level classification.

4. The complex business project review system based on multi-dimensional historical data according to claim 3 is characterized in that: When the initial risk score corresponding to the business project review feature data exceeds a preset risk score threshold, the business project review feature data is subjected to a secondary risk level classification, and the following operations are performed: Retrieve the number of associated business nodes and the probability of abnormal occurrence corresponding to the associated business nodes; Obtaining a standard deviation of node anomaly probability based on the number of associated service nodes and the anomaly occurrence probability corresponding to the associated service nodes; Retrieve the risk probability corresponding to each business process link; The risk probability of each business process link corresponding to the risk probability threshold is compared with the preset risk probability threshold, wherein the preset risk probability threshold value is 0.4; Screening out business process links whose risk occurrence probability is not lower than the preset risk probability threshold as target business process links; Determine whether non-target business process links are interspersed between target business process links, and obtain each group of target business process links interspersed with non-target business process links; Set the risk impact coefficient based on each group of target business process links that are mixed with non-target business process links; The risk impact coefficient is combined with the node abnormality probability standard deviation to obtain the secondary risk determination parameter; Comparing the secondary risk determination parameter with a preset risk determination parameter threshold; When the secondary risk determination parameter is lower than the preset risk determination parameter threshold, the business project review task is determined to be of medium risk level; When the secondary risk determination parameter is not lower than a preset risk determination parameter threshold, the business project review task is determined to be at a high risk level.

5. The complex business project review system based on multi-dimensional historical data according to claim 2 is characterized in that: Prioritize and intelligently assign tasks by: Prioritize complex business project review tasks based on the business project review risk assessment results, combining risk level factors, task complexity factors, and urgency factors to determine task priority; Dynamically assign tasks based on task complexity, reviewer capabilities, work status, and task priority to optimize resource allocation.

6. The complex business project review system based on multi-dimensional historical data according to claim 3 is characterized in that: Preprocess multi-source data for business project review and perform the following operations: Clean the multi-source data of business project audits, remove noise irrelevant to complex business project audits, and process missing values ​​and outliers in the multi-source data of business project audits; The multi-source data of business project review is converted into a unified data format, the dimensional differences in the multi-source data of business project review are removed, and standardized multi-source data of business project review is formed.

7. The complex business project review system based on multi-dimensional historical data according to claim 4 is characterized in that: Preprocess multi-source data for business project review and perform the following operations: Integrate multi-source data of business project review, merge multi-source data of business project review from different sources into a unified data view, and store the integrated multi-source data of business project review; Perform feature extraction on multi-source data of business project review, extract features related to complex business project review from the multi-source data of business project review, and determine business project review feature data.

8. The complex business project review system based on multi-dimensional historical data according to claim 5 is characterized in that: Utilize machine learning algorithms to build a business project review risk assessment model that performs the following operations: Collecting historical data on business project reviews and dividing the historical data into a training set and a test set; Based on machine learning algorithms, a training set is used to train the machine learning model, allowing the machine learning model to autonomously learn complex business project audit behaviors from the training set, identify abnormal patterns and potential risks in audit tasks, and determine a business project audit risk assessment model based on machine learning; Input the test set into the business project review risk assessment model, test the business project review risk assessment model based on the test set, evaluate the performance of the business project review risk assessment model, and determine the model test evaluation results; The business project review risk assessment model is adjusted and optimized based on the model test evaluation results to determine the optimal business project review risk assessment model.

9. The complex business project review system based on multi-dimensional historical data according to claim 8 is characterized in that: To evaluate the performance of the business project review risk assessment model, do the following: Use the test set to test the business project audit risk assessment model, and judge whether the business project audit risk assessment model can achieve the expected effect of identifying abnormal patterns and potential risks in audit tasks based on the accuracy, precision and recall rate; When the business project audit risk assessment model cannot achieve the expected effect of identifying abnormal patterns and potential risks in the audit task, the parameters of the business project audit risk assessment model shall be adjusted and optimized until the business project audit risk assessment model can achieve the expected effect of identifying abnormal patterns and potential risks in the audit task, and the optimal business project audit risk assessment model shall be determined.

10. The complex business project review system based on multi-dimensional historical data according to claim 7, characterized in that: Use data visualization tools to display audit progress, risk distribution, and resource usage, and perform the following actions: Use data visualization tools to intuitively display audit progress, risk distribution, and resource usage through a variety of charts and interactive dashboards; Regularly generate complex business project audit reports based on audit progress, risk distribution, and resource usage to provide management with detailed information on task allocation and resource utilization.

Images