Business process data management method and system based on artificial intelligence

Abstract

The embodiment of the invention provides a business process data management method and system based on artificial intelligence, and the method comprises the steps: analyzing the priori key and other priori business processes of a to-be-triggered software repair application business process; respectively carrying out business process derivation on the first prior software repair application business process of the to-be-triggered software repair application business process, the target business process path data and the other business process path data; the first derivation data and the second derivation data which reflect the confidence coefficient of skipping to the next software repair application business process under different conditions are generated, and the deviation confidence coefficient between the first derivation data and the second derivation data is compared with the preset threshold confidence coefficient; according to the method, the to-be-triggered software repair application business process can be accurately classified into the target software repair application business process or other software repair application business processes, so that the business process data management efficiency and accuracy are greatly improved, and the business process data can be effectively managed.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of artificial intelligence technology, and more specifically, to a business process data management method and system based on artificial intelligence. Background Art

[0002] With the rapid development of information technology, the complexity and diversity of business processes are increasing, and effective data management of business processes is particularly important. Traditional business process data management methods often rely on manual analysis and processing, which is inefficient and error-prone, and cannot meet the needs of modern enterprises for efficient and accurate data management. Especially in the field of software repair applications, the optimization and management of business processes are directly related to the quality of the software and user experience.

[0003] However, related technologies have obvious deficiencies when dealing with software repair application business processes and are unable to provide comprehensive and accurate data support for the system platform, which restricts the optimization of business processes and decision-making efficiency. Summary of the invention

[0004] In order to at least overcome the above-mentioned deficiencies in the prior art, the purpose of the embodiments of the present application is to provide a business process data management method and system based on artificial intelligence.

[0005] In a first aspect, an embodiment of the present application provides a business process data management method based on artificial intelligence, the method comprising: Determine target business process path data of a target software repair application business process based on a priori key software repair application business process of the software repair application business process to be triggered, and determine other business process path data of other software repair application business processes based on a priori other software repair application business processes of the software repair application business process to be triggered, wherein the other software repair application business processes are software repair application business processes other than the target software repair application business process; Performing business process derivation on the first prior software repair application business process of the to-be-triggered software repair application business process and the target business process path data to generate first derivation data, wherein the first derivation data is used to reflect the confidence of jumping to the next software repair application business process after terminating the to-be-triggered software repair application business process when the to-be-triggered software repair application business process is the target software repair application business process; Performing business process derivation on the first a priori software repair application business process of the to-be-triggered software repair application business process and the other business process path data to generate second derivation data, wherein the second derivation data is used to reflect the confidence of jumping to the next software repair application business process after terminating the to-be-triggered software repair application business process when the to-be-triggered software repair application business process is the other software repair application business process; When the deviation confidence between the first derivation data and the second derivation data is greater than the threshold confidence, the software repair application business process to be triggered is output as the target software repair application business process; when the deviation confidence between the first derivation data and the second derivation data is not greater than the threshold confidence, the software repair application business process to be triggered is output as the other software repair application business process.

[0006] In a possible implementation of the first aspect, determining target business process path data of a target software repair application business process according to a priori key software repair application business process of the software repair application business process to be triggered includes: Acquire a second a priori software repair application service process of the a priori critical software repair application service process, where the second a priori software repair application service process is another software repair application service process before the a priori critical software repair application service process; Performing regular conversion on the business process connection degree between the second a priori software repair application business process and the first a priori software repair application business process to generate a business process influence coefficient of the a priori key software repair application business process; The prior critical business process path data is extracted from the prior critical software repair application business process, and the target business process path data of the target software repair application business process is determined based on the business process impact coefficient of the prior critical software repair application business process and the prior critical business process path data.

[0007] In a possible implementation manner of the first aspect, before determining the target business process path data of the target software repair application business process according to the business process impact coefficient of the a priori critical software repair application business process and the a priori critical business process path data, the method further includes: Acquire a target business process participating node corresponding to the software repair application business process to be triggered, and acquire a reference business process participating node having a business process connection degree with the target business process participating node greater than a preset connection degree from a business process participating node sequence, wherein the business process participating node sequence includes a plurality of business process participating nodes; Extracting reference a priori critical software repair application business process path data from the reference business process participating node's a priori critical software repair application business process, and determining a business process impact coefficient of the reference business process participating node's a priori critical software repair application business process; The determining, based on the business process impact coefficient of the a priori critical software repair application business process and the a priori critical business process path data, the target business process path data of the target software repair application business process comprises: According to the business process influence coefficient of the a priori key software repair application business process of the software repair application business process to be triggered, the a priori key business process path data of the software repair application business process to be triggered are merged to generate the first business process path data of the target software repair application business process; According to the business process influence coefficient of the a priori critical software repair application business process of the reference business process participating node, the a priori critical business process path data of the reference business process participating node are integrated to generate the second business process path data of the target software repair application business process; Integrating the first business process path data and the second business process path data to generate target business process path data of the target software repair application business process; In a possible implementation of the first aspect, the step of acquiring, from the business process participating node sequence, a reference business process participating node having a business process connectivity with the target business process participating node greater than a preset connectivity, includes: Acquire a database of a sequence of nodes participating in a business process, wherein the database records information of all nodes that have participated in the software repair application business process, and the node information includes an identifier of the node participating in the business process, the software repair application business process that it has participated in historically, and performance and evaluation data of the software repair application business process; For each candidate business process participating node in the business process participating node sequence, analyzing the candidate software repair application business processes that the business process participating node has historically participated in and that have an intersection with the target business process participating node; Obtaining the number of software repair application business processes and the similarity of participation types in which each candidate business process participating node and the target business process participating node jointly participate in the software repair application business process data of the candidate software repair application business process, and obtaining first evaluation data; Obtaining the sequence and dependency relationship between each candidate business process participating node and the target business process participating node in the jointly participating candidate software repair application business process to obtain second evaluation data; Obtaining the time proximity of the candidate software repair application business process jointly participated by each candidate business process participating node and the target business process participating node to obtain third evaluation data; Obtaining the similarity of performance evaluation between each candidate business process participating node and the target business process participating node in the jointly participating candidate software repair application business process, to obtain fourth evaluation data; According to the first evaluation data, the second evaluation data, the third evaluation data and the fourth evaluation data, corresponding evaluation weights are set and weighted fusion is performed to generate a business process connection degree between each candidate business process participating node and a target business process participating node; The business process connection degree is compared with a preset connection degree, and when the business process connection degree is greater than the preset connection degree, the corresponding candidate business process participating node is output as the reference business process participating node.

[0008] In a possible implementation manner of the first aspect, before regularizing the business process connection degree between the second a priori software repair application business process and the first a priori software repair application business process to generate the business process impact coefficient of the a priori key software repair application business process, the method further includes: For each of the second a priori software repair application service processes, obtaining second a priori service process path data from the second a priori software repair application service process; Obtaining first a priori business process path data from the first a priori software repair application business process of the to-be-triggered software repair application business process; The business process connection degree between the first a priori business process path data and the second a priori business process path data is output as the business process connection degree between the second a priori software repair application business process and the first a priori software repair application business process.

[0009] In a possible implementation manner of the first aspect, before determining other service process path data of other software repair application service processes based on the prior other software repair application service processes of the to-be-triggered software repair application service process, the method further includes: Decomposing the priori software repair application service process of the to-be-triggered software repair application service process into time periods to generate a plurality of priori software repair application service process time periods, each of the priori software repair application service process time periods including at least one of the priori software repair application service processes; When the final software repair application service process in the priori software repair application service process period is the other software repair application service process, the final software repair application service process in the priori software repair application service process period is output as the priori other software repair application service process of the software repair application service process to be triggered.

[0010] In a possible implementation manner of the first aspect, the performing business process derivation on the first a priori software repair application business process of the to-be-triggered software repair application business process and the target business process path data to generate first derivation data includes: Extracting first a priori business process path data from the first a priori software repair application business process; Determine a first business process path feature according to the first priori business process path data and the target business process path data; Calling a first business process inference network, wherein the first business process inference network includes a first graph self-attention unit and a first prediction unit; Using the first graph self-attention unit to perform graph self-attention conversion on the first business process path feature to generate a first graph self-attention conversion vector; Using the first prediction unit to perform business process derivation on the first graph self-attention conversion vector to generate the first derivation data; In a possible implementation manner of the first aspect, determining the first business process path feature according to the first priori business process path data and the target business process path data includes: Fusion of the first priori business process path data and the target business process path data to generate the first business process path feature; Performing weighted processing on the first priori business process path data and the target business process path data to generate the first business process path feature; The target business process participating node vector corresponding to the to-be-triggered software repair application business process is obtained, and the first priori business process path data, the target business process path data and the target business process participating node vector are integrated to generate the first business process path feature.

[0011] In a possible implementation manner of the first aspect, before calling the first service process to derive the network, the method further includes: configuring first sample business process path data and sample business process derivation data of the first sample business process path data according to the a priori critical software repair application business process; Extracting first sample business process path features from the first sample business process path data; Using the graph self-attention unit of the basic business process derivation network, perform graph self-attention conversion on the first sample business process path feature to generate a sample graph self-attention conversion vector; Using the prediction unit of the basic business process derivation network, the business process is deduced for the sample graph self-attention conversion vector to generate training derivation data; Training the basic business process derivation network according to the training derivation data and the sample business process derivation data, and outputting the network parameter information of the trained basic business process derivation network as the network parameter information of the first business process derivation network; The configuring the first sample business process path data and the sample business process derivation data of the first sample business process path data according to the prior critical software repair application business process includes: Merging a second a priori software repair application business process of the a priori critical software repair application business process with the a priori critical software repair application business process to generate first sample business process path data; Determine a first progress point when the a priori critical software repair application service process is terminated and a second progress point when the next software repair application service process of the a priori critical software repair application service process is triggered; When the difference between the second progress point and the first progress point is greater than a set difference, the sample business process derivation data of the first sample business process path data is output as a first sample business process derivation state; when the difference between the second progress point and the first progress point is not greater than the set difference, the sample business process derivation data of the first sample business process path data is output as a second sample business process derivation state; In a possible implementation of the first aspect, the step of using a graph self-attention unit of a basic business process derivation network to perform graph self-attention conversion on the first sample business process path feature to generate a sample graph self-attention conversion vector includes: Extracting a multidimensional feature vector of the first sample business process path feature, and adding a position code to each feature vector in the multidimensional feature vector to indicate the position of the feature vector in the first sample business process path feature; Calculating the attention score between each input feature vector and all other feature vectors in the multi-dimensional feature vector by the graph self-attention unit; Based on the calculated attention score, a preset mask is applied to shield invalid feature vectors, and a scaling factor is applied to adjust the distribution of the attention score. Then, according to the attention score, each remaining feature vector after the shielding process is weighted summed to generate the sample image self-attention conversion vector.

[0012] In a possible implementation manner of the first aspect, the performing business process derivation on the first a priori software repair application business process of the to-be-triggered software repair application business process and the other business process path data to generate second derivation data includes: Extracting first a priori business process path data from the first a priori software repair application business process; generating a second business process path feature according to the first a priori business process path data and the other business process path data; Calling a second business process inference network, wherein the second business process inference network includes a second graph self-attention unit and a second prediction unit; Using the second graph self-attention unit to perform graph self-attention conversion on the second business process path feature to generate a second graph self-attention conversion vector; The second prediction unit is used to derive the business process of the second graph self-attention conversion vector to generate the second derivation data.

[0013] In a possible implementation manner of the first aspect, after outputting the to-be-triggered software repair application service process as the target software repair application service process, the method further includes: Outputting the a priori software repair application service process number of the to-be-triggered software repair application service process as a first a priori software repair application service process number; Outputting the a priori software repair application business process number of the a priori critical software repair application business process as a second a priori software repair application business process number; When the difference between the first a priori software repair application business process number and the second a priori software repair application business process number is less than a preset value, the to-be-triggered software repair application business process is migrated from the target software repair application business process to the other software repair application business process.

[0014] In the second aspect, the embodiments of the present application also provide an artificial intelligence-based business process data management system, which includes a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions, and the machine-executable instructions are loaded and executed by the processor to implement the artificial intelligence-based business process data management method in any possible implementation manner in the first aspect.

[0015] According to any of the above aspects, the embodiment of the present application can accurately determine the target business process path data and other business process path data by analyzing the prior key and the prior other business processes of the software repair application business process to be triggered, thereby providing an accurate data basis for the subsequent business process derivation. Secondly, by deducing the first prior software repair application business process and the target business process path data and the other business process path data of the software repair application business process to be triggered, the first derivation data and the second derivation data reflecting the confidence of jumping to the next software repair application business process under different circumstances are generated, which provides an important basis for judging the nature of the software repair application business process to be triggered. Finally, by comparing the deviation confidence between the first derivation data and the second derivation data with the preset threshold confidence, the software repair application business process to be triggered can be accurately classified as the target software repair application business process or other software repair application business process. This classification method is not only highly accurate, but also can be completed automatically, which greatly improves the efficiency and accuracy of business process data management. Therefore, the business process data can be effectively managed and the intelligent level of data processing can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required to be used in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be extracted in combination with these drawings without creative work.

[0017] Figure 1 A flowchart of a business process data management method based on artificial intelligence provided by an embodiment of the present invention; Figure 2 A schematic block diagram of the structure of an artificial intelligence-based business process data management system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following description is intended to enable one of ordinary skill in the art to implement and incorporate the present invention, and the description is provided in the context of a specific application scenario and its requirements. It will be apparent to one of ordinary skill in the art that various changes may be made to the disclosed embodiments, and that the general principles defined in the present invention may be applied to other embodiments and application scenarios without departing from the principles and scope of the present invention. Therefore, the present invention is not limited to the described embodiments, but should be given the broadest scope consistent with the claims.

[0019] Figure 1This is a flow chart of a business process data management method based on artificial intelligence provided by an embodiment of the present invention. The business process data management method based on artificial intelligence is introduced in detail below.

[0020] Step S110, determining the target business process path data of the target software repair application business process based on the prior key software repair application business process of the software repair application business process to be triggered, and determining other business process path data of other software repair application business processes based on the prior other software repair application business processes of the software repair application business process to be triggered, wherein the other software repair application business processes are software repair application business processes other than the target software repair application business process.

[0021] In detail, the software repair application business process to be triggered may refer to a software repair task or process that currently needs to be determined to belong to (i.e., to which main repair process it belongs). For example, assuming that a software system has a problem of database performance degradation, then "database performance optimization" may be a software repair application business process to be triggered.

[0022] The a priori critical software repair application business process may refer to a software repair task or process that is directly logically associated with the software repair application business process to be triggered and is usually executed before it. This association may be based on historical data, expert knowledge or business logic. For example, in the context of "database performance optimization", "performance evaluation after system upgrade" may be a a priori critical software repair application business process, because performance evaluation is often required after the system upgrade to determine whether further performance optimization is needed.

[0023] The target software repair application business process may be the main repair process or business line to which the software repair application business process to be triggered is judged to belong. For example, if "database performance optimization" is judged to be part of the "system post-upgrade tuning" process, then it is a target software repair application business process in the process. The target business process path data may be data describing how the target software repair application business process should be executed, including execution steps, required resources, expected results, etc. For example, for "database performance optimization", the target business process path data may include specific optimization steps (such as query optimization, index adjustment, etc.), required tools or resources (such as performance monitoring tools, database administrators, etc.), and expected performance improvement indicators.

[0024] The a priori other software repair application business process may be other software repair tasks or processes associated with the software repair application business process to be triggered, in addition to the a priori key software repair application business process and the target software repair application business process. For example, when determining the attribution of "database performance optimization", "network troubleshooting" may be regarded as an a priori other software repair application business process, because although it is related to database performance, it does not directly belong to the "post-system upgrade tuning" process. The other business process path data may be data that describes how the a priori other software repair application business process is executed. For example, for "network troubleshooting", its business process path data may include troubleshooting steps, required network monitoring tools, and expected fault resolution time, etc.

[0025] Therefore, in this embodiment, in an actual software repair application scenario, the server is responsible for managing a large software repair project, which includes multiple software repair application business processes, and these software repair application business processes are performed in a certain logical order to ensure the efficiency and accuracy of software repair. When a new software repair application business process to be triggered appears, the server needs to determine which main repair process (target software repair application business process) the software repair application business process to be triggered belongs to, so as to allocate corresponding resources and policies for it.

[0026] For example, suppose that the software repair application business process to be triggered is "database performance optimization". The server will first check the historical records to find the prior critical software repair application business process related to the software repair application business process to be triggered, such as "performance evaluation after system upgrade". The prior critical software repair application business process usually has a direct logical association with the software repair application business process to be triggered because it involves performance issues that may be caused after the system upgrade. The server can determine the target business process path data of "performance evaluation after system upgrade", and these target business process path data include the execution steps, required resources, expected results, etc. of the prior critical software repair application business process. At the same time, the server will also determine other prior software repair application business processes related to "database performance optimization", such as "network troubleshooting", etc., and extract their other business process path data.

[0027] Step S120, performing business process derivation on the first prior software repair application business process of the software repair application business process to be triggered and the target business process path data to generate first derivation data, wherein the first derivation data is used to reflect the confidence of jumping to the next software repair application business process after ending the software repair application business process to be triggered when the software repair application business process to be triggered is the target software repair application business process.

[0028] In detail, the first a priori software repair application business process may be a software repair task or process that is executed before the software repair application business process to be triggered and is directly related to it. This may be one of the a priori key software repair application business processes. For example, before "database performance optimization", there may be a "database performance monitoring" process for collecting performance data for optimization. This "database performance monitoring" can be regarded as the first a priori software repair application business process.

[0029] In this embodiment, the server can analyze the first prior software repair application business process (which can be a previous "database performance monitoring" process) of the software repair application business process to be triggered, "database performance optimization", and derive the business process in combination with the target business process path data of "system post-upgrade performance evaluation". The derivation process can involve simulating the execution path, predicting resource requirements, evaluating potential risks, etc., and finally generating the first derivation data, which reflects the confidence of jumping to the next process (such as "performance tuning result feedback") after executing the target software repair application business process if "database performance optimization" is part of the target software repair application business process.

[0030] Step S130, performing business process deduction on the first prior software repair application business process of the software repair application business process to be triggered and the other business process path data to generate second derivation data, wherein the second derivation data is used to reflect the confidence of jumping to the next software repair application business process after ending the software repair application business process to be triggered when the software repair application business process to be triggered is the other software repair application business process.

[0031] Similarly, the server will also perform business process deduction on the first prior software repair application business process of "database performance optimization" and other business process path data of "network troubleshooting" to generate second deduction data. The second deduction data reflects the confidence of jumping to the next process after executing the process if "database performance optimization" is actually part of the "network troubleshooting" process.

[0032] Step S140, when the deviation confidence between the first derivation data and the second derivation data is greater than the threshold confidence, the software repair application business process to be triggered is output as the target software repair application business process; when the deviation confidence between the first derivation data and the second derivation data is not greater than the threshold confidence, the software repair application business process to be triggered is output as the other software repair application business process.

[0033] In this embodiment, the server can compare the deviation confidence between the first derived data and the second derived data. If the confidence of the first derived data is significantly higher than that of the second derived data (exceeding a preset threshold confidence), then the server can consider that "database performance optimization" is part of the "system post-upgrade performance evaluation" process, that is, the target software repair application business process. The server can mark the software repair application business process to be triggered as the target software repair application business process and allocate corresponding resources and policies to it. On the contrary, if the deviation confidence is not large or the confidence of the second derived data is higher, the server will mark the software repair application business process to be triggered as other software repair application business processes and perform corresponding processing.

[0034] As a result, the server can intelligently determine the main repair process to which the software repair application business process to be triggered belongs, thereby ensuring efficient execution of the software repair project.

[0035] Based on the above steps, the embodiment of the present application can accurately determine the target business process path data and other business process path data by analyzing the prior key and prior other business processes of the software repair application business process to be triggered, providing an accurate data basis for the subsequent business process derivation. Secondly, by deducing the first prior software repair application business process and the target business process path data and other business process path data of the software repair application business process to be triggered, the first derivation data and the second derivation data reflecting the confidence of jumping to the next software repair application business process under different circumstances are generated, which provides an important basis for judging the nature of the software repair application business process to be triggered. Finally, by comparing the deviation confidence between the first derivation data and the second derivation data with the preset threshold confidence, the software repair application business process to be triggered can be accurately classified as the target software repair application business process or other software repair application business process. This classification method is not only highly accurate, but also can be completed automatically, greatly improving the efficiency and accuracy of business process data management. Therefore, the business process data can be effectively managed and the intelligent level of data processing can be improved.

[0036] In a possible implementation, step S110 may include: Step S111 , obtaining a second a priori software repair application service process of the a priori critical software repair application service process, where the second a priori software repair application service process is another software repair application service process before the a priori critical software repair application service process.

[0037] Step S112, performing regular conversion on the business process connection degree between the second a priori software repair application business process and the first a priori software repair application business process to generate a business process influence coefficient of the a priori key software repair application business process.

[0038] Step S113, extracting the a priori critical business process path data from the a priori critical software repair application business process, and determining the target business process path data of the target software repair application business process based on the business process impact coefficient of the a priori critical software repair application business process and the a priori critical business process path data.

[0039] In this embodiment, assuming that the software repair application business process to be triggered is "database performance optimization", the server first searches the historical records for a priori key software repair application business process that is directly logically associated with "database performance optimization", such as "system performance evaluation after upgrade".

[0040] Next, the server may obtain a second priori software repair application service process before the “system post-upgrade performance evaluation”, which may be “system upgrade deployment” or “new version testing”.

[0041] The server can analyze the business logic connection between "system upgrade deployment" (second prior software repair application business process) and "system post-upgrade performance evaluation" (prior key software repair application business process), which can include analyzing the execution order, resource dependency, expected results, etc. between them.

[0042] Through a series of regular transformations, the server calculates the business process connection degree between the two processes and converts it into the business process impact coefficient of the a priori critical software repair application business process (i.e., "system post-upgrade performance evaluation"). This impact coefficient reflects the importance of "system post-upgrade performance evaluation" in the entire software repair project and its impact on subsequent processes.

[0043] The server extracts the prior critical business process path data from the prior critical software repair application business process of "system post-upgrade performance evaluation", which may include the detailed execution steps, required resources, expected results, historical performance, etc. of the process.

[0044] Combining the previously calculated business process impact coefficient and the extracted a priori key business process path data, the server generates target business process path data for the target software repair application business process (i.e., "database performance optimization"). This path data provides the server with detailed guidance on how to execute the "database performance optimization" process, including resource allocation, execution order, potential risks, etc.

[0045] Through this process, the server can accurately determine the main repair process to which the software repair application business process to be triggered belongs, and formulate a suitable execution strategy for it, thereby ensuring the efficient execution of the entire software repair project.

[0046] In a possible implementation, before step S113, the method further includes: Step A110, obtain the target business process participating node corresponding to the software repair application business process to be triggered, and obtain a reference business process participating node whose business process connection degree with the target business process participating node is greater than a preset connection degree from the business process participating node sequence, wherein the business process participating node sequence includes multiple business process participating nodes.

[0047] Step A120, extracting reference a priori critical software repair application business process path data from the reference business process participating node's a priori critical software repair application business process, and determining the business process impact coefficient of the reference business process participating node's a priori critical software repair application business process.

[0048] In this embodiment, it is assumed that the software repair application business process to be triggered is "database performance optimization". The server first determines the target business process participating nodes of the software repair application business process to be triggered, and these target business process participating nodes can be specific teams or personnel responsible for database maintenance, performance tuning or system management.

[0049] In addition, the server maintains a database of business process participating node sequences, which records information of all nodes that have participated in the software repair application business process, including node identification, historically participated software repair application business processes, and performance and evaluation data in those processes.

[0050] The server can traverse the sequence of business process participating nodes to find other nodes that have had similar participation experience in the history with the target business process participating node of "database performance optimization". These nodes may have similar roles or responsibilities with the target business process participating node in the software repair application business process in which they have previously participated together, so they may provide valuable reference information for the target business process.

[0051] The server can calculate the business process connectivity between each candidate node and the target business process participating node, and the connectivity is based on multiple factors, such as the number of jointly participating software repair application business processes, the similarity of participation types, the sequence and dependencies in the jointly participating processes, and the time proximity and similarity of performance evaluations.

[0052] When the connectivity of a candidate node is greater than the preset connectivity threshold, it will be regarded as a reference business process participating node.

[0053] For each reference business process participating node, the server can view the prior critical software repair application business processes in which they previously participated, and extract relevant prior critical business process path data from them, which describes the execution steps, required resources, historical results, etc. of these processes.

[0054] The server also calculates the business process impact coefficient of the a priori critical software repair application business processes of these reference business process participating nodes, and the business process impact coefficient reflects the degree of influence of these a priori critical software repair application business processes on the current behavior of the reference business process participating nodes.

[0055] Step S113 includes: Step S1131, based on the business process influence coefficient of the a priori key software repair application business process of the software repair application business process to be triggered, the a priori key business process path data of the software repair application business process to be triggered is integrated to generate the first business process path data of the target software repair application business process.

[0056] Step S1132, based on the business process impact coefficient of the a priori critical software repair application business process of the reference business process participating node, the a priori critical business process path data of the reference business process participating node is integrated to generate second business process path data of the target software repair application business process.

[0057] Step S1133: Integrate the first business process path data and the second business process path data to generate target business process path data of the target software repair application business process.

[0058] In this embodiment, after having the prior key business process path data of the software repair application business process to be triggered and the relevant information of the reference business process participating nodes, the server starts to generate the target business process path data: The server integrates and adjusts the a priori key business process path data of the software repair application business process to be triggered based on the business process influence coefficient of the a priori key software repair application business process of the software repair application business process to be triggered. This process may include weighting certain steps in the path data according to the influence coefficient, adjusting resource allocation or optimizing the execution order, etc.

[0059] For each reference business process participating node, the server extracts relevant information from the corresponding prior key business process path data based on the business process impact coefficient of its prior key software repair application business process, and integrates and adjusts it to generate second business process path data. These second business process path data provide reference paths obtained from similar experiences.

[0060] Finally, the server integrates the first business process path data and all the second business process path data, which may include comparing the similarities of different paths, identifying commonalities and differences, and merging and optimizing paths according to preset rules or policies. The target business process path data finally generated will provide comprehensive and targeted execution guidance for the "database performance optimization" software repair application business process to be triggered.

[0061] In a possible implementation, step A110 may include: Step A111, obtain a database of the business process participating node sequence, the database records all node information that has participated in the software repair application business process, the node information includes the business process participating node identification, historically participated software repair application business process, software repair application business process performance and evaluation data.

[0062] In this embodiment, the server first accesses a database that stores information about all nodes that have participated in the software repair application business process. The database records in detail the identification of each business process participating node (such as team ID, person name, etc.), a list of software repair application business processes they have historically participated in, and performance and evaluation data in these processes.

[0063] Step A112: for each candidate business process participating node in the business process participating node sequence, analyzing the candidate software repair application business processes that the business process participating node has historically participated in and that have an intersection with the target business process participating node.

[0064] For each candidate business process participating node in the business process participating node sequence, the server can analyze the list of software repair application business processes that the candidate business process participating node has historically participated in, and find out the candidate software repair application business processes that have intersections with the target business process participating node. These intersections indicate that the two nodes may have similar responsibilities or experiences in previous software repair projects.

[0065] Step A113, obtaining the number of software repair application business processes and the similarity of participation types in which each candidate business process participating node and the target business process participating node participate in the software repair application business process data of the candidate software repair application business process, and obtaining first evaluation data.

[0066] Step A114, obtaining the sequence and dependency relationship between each candidate business process participating node and the target business process participating node in the jointly participating candidate software repair application business process, and obtaining second evaluation data.

[0067] Step A115 , obtaining the time proximity between each candidate business process participating node and the candidate software repair application business process jointly participated by the target business process participating node, and obtaining third evaluation data.

[0068] Step A116, obtaining the similarity of the performance evaluation of each candidate business process participating node and the target business process participating node in the jointly participating candidate software repair application business process, and obtaining fourth evaluation data.

[0069] In this embodiment, the server can count the number of software repair application business processes that each candidate business process participating node and the target business process participating node jointly participate in, and evaluate the similarity of the participation types of these processes (such as responsible modules, scope of responsibilities, etc.). For example, if two nodes often participate in database performance optimization projects, then the similarity of their participation types will be very high.

[0070] The server can further analyze the sequence and dependency relationship between each candidate business process participating node and the target business process participating node in the jointly participating candidate software repair application business process, including determining which node usually starts working first and whether there is a dependency between their work. This information helps to understand the collaboration mode of the two nodes in the project.

[0071] The server also calculates the temporal proximity of each candidate business process participating node to the software repair application business process that the target business process participating node jointly participated in. If two nodes have recently jointly participated in many projects, their work experience and collaboration methods may be more similar and fresh.

[0072] The server can access the performance and evaluation data stored in the database and compare the performance evaluation of each candidate business process participating node with the target business process participating node in the software repair application business process they jointly participate in. If their evaluations are highly similar (such as both received good or bad reviews), then their work styles and capabilities may be more matched.

[0073] Step A117, according to the first evaluation data, the second evaluation data, the third evaluation data and the fourth evaluation data, set corresponding evaluation weights and perform weighted fusion to generate the business process connection degree between each candidate business process participating node and the target business process participating node.

[0074] Step A118, comparing the business process connection degree with a preset connection degree, and when the business process connection degree is greater than the preset connection degree, outputting the corresponding candidate business process participating node as the reference business process participating node.

[0075] In this embodiment, the server can set corresponding evaluation weights for the first evaluation data, the second evaluation data, the third evaluation data, and the fourth evaluation data according to a preset rule or strategy. Then, the first evaluation data, the second evaluation data, the third evaluation data, and the fourth evaluation data can be weighted and fused to generate a comprehensive business process connection degree, which reflects the overall connection degree between the candidate business process participating node and the target business process participating node in the software repair application business process.

[0076] Finally, the server can compare the calculated business process connectivity with the preset connectivity threshold. If the business process connectivity of a candidate business process participating node exceeds the threshold, it will be identified as a reference business process participating node, and its historical experience and collaboration model will be used to guide the execution of the target software repair application business process.

[0077] In a possible implementation, before step S112, the method further includes: Step B110: for each of the second a priori software repair application service processes, obtain second a priori service process path data from the second a priori software repair application service process.

[0078] Step B120: obtaining first a priori business process path data from the first a priori software repair application business process of the software repair application business process to be triggered.

[0079] Step B130: outputting the business process connection degree between the first a priori business process path data and the second a priori business process path data as the business process connection degree between the second a priori software repair application business process and the first a priori software repair application business process.

[0080] In this embodiment, the server first searches the history records to find the prior critical software repair application business process associated with the software repair application business process to be triggered, such as "system post-upgrade performance evaluation". Then, it further traces back to the second prior software repair application business process before "system post-upgrade performance evaluation", such as "system upgrade deployment".

[0081] For each second-priori software repair application business process (such as "system upgrade deployment"), the server can extract relevant business process path data from its historical records. This data describes in detail the execution steps, required resources, expected results, etc. of the process, which is called second-priori business process path data.

[0082] Next, the server can return to the direct predecessor of the software repair application business process to be triggered (such as "database performance optimization"), that is, the first prior software repair application business process (such as "database performance monitoring"). From this process, the server can extract relevant business process path data, including execution steps, resource requirements, expected output, etc., which are called first prior business process path data.

[0083] The server may analyze the similarity and correlation between the first a priori business process path data and the second a priori business process path data, and the process may involve multiple aspects: Step matching: The server can compare whether the execution steps in the two path data have overlapping or similar parts, and whether the execution order of these steps is reasonable.

[0084] Resource requirement consistency: The server can check whether the required resources in the two path data overlap and whether the allocation of these resources is reasonable.

[0085] Expected result association: The server can evaluate whether the expected results of two path data are logically related, and whether the output of one process can be used as the input of another process.

[0086] Based on the above analysis, the server can calculate a business process connectivity degree, which reflects the logical association degree and mutual dependence between the second a priori software repair application business process and the first a priori software repair application business process.

[0087] Finally, the server can use the calculated business process connectivity as an input factor for generating a business process impact coefficient for a priori critical software repair application business processes (such as "performance evaluation after system upgrade"), and the business process impact coefficient will be used in subsequent decision-making processes, such as determining the target business process path data of the target software repair application business process.

[0088] In this way, the server can more accurately evaluate the correlation and dependency between different software repair application business processes, thereby providing more precise and effective management for the entire software repair project.

[0089] In a possible implementation manner, before determining other service process path data of other software repair application service processes based on the prior other software repair application service processes of the to-be-triggered software repair application service process, the method further includes: Step C110, time-segmenting the priori software repair application business process of the to-be-triggered software repair application business process to generate a plurality of priori software repair application business process time periods, each of which includes at least one priori software repair application business process.

[0090] Step C120, when the final software repair application service process in the priori software repair application service process period is the other software repair application service process, the final software repair application service process in the priori software repair application service process period is output as the priori other software repair application service process of the software repair application service process to be triggered.

[0091] In this embodiment, in an actual software repair project management scenario, before determining other business process path data of other software repair application business processes related to the software repair application business process to be triggered, the server will first time period-decompose the prior software repair application business process of the software repair application business process to be triggered to identify possible prior other software repair application business processes.

[0092] In detail, let's assume that the software repair application business process to be triggered is "database performance optimization". The server will first review the history of this process and find out all the prior software repair application business processes related to it. These prior software repair application business processes cover multiple aspects from system upgrade to database monitoring.

[0093] The server can then perform time period decomposition on these prior software repair application business processes. The purpose of time period decomposition is to divide the continuous related processes on the timeline into several time periods, and the processes in each time period have a relatively close relationship in logic or time. For example, the server can divide processes such as system upgrade, new version testing, and preliminary performance evaluation into one time period, because they are usually a series of operations performed immediately after the system upgrade.

[0094] After the time period is decomposed, the server can obtain multiple prior software repair application business process time periods. Each time period contains one or more prior software repair application business processes, and these processes are related to each other in logic and time.

[0095] For each priori software repair application business process period, the server can determine the final software repair application business process within the period. The final process is usually the last process executed logically or temporally within the period, which marks the completion of a series of related operations within the period.

[0096] The server can check the final software repair application business process in each prior software repair application business process period to determine whether it can be regarded as a priori other software repair application business process of the software repair application business process to be triggered, where "other" refers to other related processes except the main repair process (i.e., the target software repair application business process) to which the software repair application business process to be triggered belongs.

[0097] For example, if the final process within a certain period is "network troubleshooting", and "network troubleshooting" and "database performance optimization" are not directly related logically, or they belong to different main repair processes, then the server will identify "network troubleshooting" as a priori other software repair application business process to trigger the software repair application business process (i.e., "database performance optimization").

[0098] Once the server determines the prior other software repair application business processes of the to-be-triggered software repair application business process, it will use these processes as objects for subsequent analysis. In the next step, the server can further extract other business process path data of these prior other software repair application business processes to perform more in-depth business process derivation and analysis.

[0099] In a possible implementation, step S120 may include: Step S121: extracting first a priori business process path data from the first a priori software repair application business process.

[0100] Step S122: determining a first business process path feature according to the first priori business process path data and the target business process path data.

[0101] Step S123, calling a first business process inference network, wherein the first business process inference network includes a first graph self-attention unit and a first prediction unit.

[0102] Step S124: Use the first graph self-attention unit to perform graph self-attention conversion on the first business process path feature to generate a first graph self-attention conversion vector.

[0103] Step S125: Utilize the first prediction unit to perform business process derivation on the first graph self-attention conversion vector to generate the first derivation data.

[0104] In an actual software repair project management scenario, the server can deduce the business process based on the software repair application business process to be triggered and its related prior information to predict the next possible business process path.

[0105] In detail, let's continue to assume that the software repair application business process to be triggered is "database performance optimization", and the first a priori software repair application business process directly related to it is "database performance monitoring". The server will first extract relevant business process path data from the "database performance monitoring" process. These business process path data may include specific monitoring steps, tools used, collected data types, and monitoring results, etc. These business process path data are called first a priori business process path data.

[0106] The server then combines the first prior business process path data and the target business process path data to determine the first business process path feature. The target business process path data is usually extracted based on the prior critical software repair application business process (such as "system upgrade post-performance evaluation"), and describes the execution path and expected results of the prior critical software repair application business process.

[0107] In this process, the server can compare and analyze the similarities and differences between the first prior business process path data and the target business process path data. For example, it can be checked whether the performance data collected in "Database Performance Monitoring" is consistent with the expected database performance optimization goals in the target business process path, or whether it reveals specific aspects that need to be optimized. Based on these analyses, the server can generate a feature vector that combines the first prior business process path data and the target business process path data, namely, the first business process path feature.

[0108] The server then calls a pre-trained first business process inference network to perform business process inference. The first business process inference network is a deep learning model that includes a first graph self-attention unit and a first prediction unit. The graph self-attention unit is used to capture the complex dependencies and structural information in the business process path data, while the prediction unit generates inference results based on this information.

[0109] In the first business process inference network, the server first inputs the first business process path feature into the first graph self-attention unit. The graph self-attention unit assigns different weights according to the relevance and importance between different parts in the feature vector. Specifically, the attention score between each element in the feature vector and other elements can be calculated, and then the feature vector is weighted and summed according to these scores to generate a new feature representation, namely the first graph self-attention transformation vector, which captures the key information and dependencies in the business process path data.

[0110] Finally, the server can input the first graph self-attention conversion vector into the first prediction unit. The first prediction unit predicts the next possible business process path based on the first graph self-attention conversion vector. Various machine learning algorithms (such as neural networks) can be used to generate derivation results. These results reflect the next software repair application business process that is most likely to jump to after the software repair application business process to be triggered (such as "database performance optimization") ends and its confidence level, given the first prior business process path data and the target business process path data. These derivation results are called first derivation data, which provide the server with predictions and guidance on the subsequent business process path.

[0111] In a possible implementation, step S122 includes: A. Fusing the first priori business process path data and the target business process path data to generate the first business process path feature.

[0112] For example, suppose the software repair application business process to be triggered is "database performance optimization", and its first a priori software repair application business process is "database performance monitoring". The server first extracts the first a priori business process path data from "database performance monitoring", which may include the monitoring time, monitoring indicators, monitoring results, etc. At the same time, the server also obtains the target business process path data related to "database performance optimization", which describes the expected steps, resource requirements, expected results, etc. of the optimization process.

[0113] In order to generate the first business process path feature, the server first merges the first prior business process path data with the target business process path data. The fusion process may include merging similar or related fields in the two data sets, and dealing with possible data format or unit inconsistencies. Through the fusion, the server obtains a comprehensive data set that includes prior monitoring information and target optimization information.

[0114] B. Perform weighted processing on the first priori business process path data and the target business process path data to generate the first business process path feature.

[0115] After fusion, the server can perform weighted processing on the first priori business process path data and the target business process path data according to the importance and relevance of the data. The weighted processing can help the server pay more attention to certain key information in subsequent analysis.

[0116] For example, if the server believes that certain monitoring indicators in "Database Performance Monitoring" (such as CPU usage, memory occupancy, etc.) have higher reference value for "Database Performance Optimization", it can assign higher weights to these indicators. Similarly, if certain steps or resource requirements in the target business process path data are critical to the optimization process, the server will increase their weights accordingly.

[0117] C. Obtain the target business process participating node vector corresponding to the software repair application business process to be triggered, and fuse the first priori business process path data, the target business process path data and the target business process participating node vector to generate the first business process path feature.

[0118] In addition to the business process path data, the server also considers the impact of participating nodes in the business process. For the "database performance optimization" business process to be triggered for software repair application, the server can determine its corresponding target business process participating nodes, which can be specific teams or individuals responsible for database maintenance, performance tuning, or system management.

[0119] The server can generate a vector representation for each target business process participating node, which may contain multi-dimensional information such as the node's historical performance, skill level, and collaboration ability. Then, the server can re-integrate this target business process participating node vector with the previously fused and weighted business process path data.

[0120] The fusion process may involve associating or mapping the node vector with specific fields in the business process path data so that both node and business process information can be considered in the analysis. By fusing the node vector, the server generates a more comprehensive first business process path feature containing multiple dimensions of information, which will serve as an important basis for the subsequent business process to derive network input.

[0121] In a possible implementation manner, before calling the first service process to derive the network, the method further includes: Step S101 : configuring first sample business process path data and sample business process derivation data of the first sample business process path data according to the a priori critical software repair application business process.

[0122] Step S102: extracting first sample business process path features from the first sample business process path data.

[0123] Step S103, using the graph self-attention unit of the basic business process deduction network, performs graph self-attention conversion on the first sample business process path feature to generate a sample graph self-attention conversion vector.

[0124] Step S104, using the prediction unit of the basic business process derivation network, the business process is deduced for the sample graph self-attention conversion vector to generate training derivation data.

[0125] Step S105 , training the basic business process derivation network according to the training derivation data and the sample business process derivation data, and outputting the network parameter information of the trained basic business process derivation network as the network parameter information of the first business process derivation network.

[0126] In this embodiment, it is still assumed that the a priori critical software repair application business process is "system post-upgrade performance evaluation". The server will first configure a series of first-instance business process path data based on this a priori critical software repair application business process. These first-instance business process path data simulate various "system post-upgrade performance evaluation" business process paths that may appear in the actual environment, including different execution steps, resource allocation, expected results, etc.

[0127] At the same time, the server can configure corresponding sample business process derivation data for each sample business process path data. These sample business process derivation data describe which next software repair application business process the system is expected to jump to under the sample business process path, as well as the confidence of jumping to the process. These sample business process derivation data will be used for subsequent network training.

[0128] The server then extracts features from each first-instance business process path data that represent the key information and attributes of the sample business process path. The extraction process may involve encoding or quantifying the steps, resource requirements, expected results, etc. in the business process path so that the machine can understand and process them. The extracted first-instance business process path features will be used as input data for subsequent network training and derivation.

[0129] The server has a pre-built basic business process inference network, which includes a graph self-attention unit that can capture complex dependencies and structural information in business process path data.

[0130] During the training process, the server can input the first sample business process path features into the graph self-attention unit of the basic business process derivation network. The graph self-attention unit will transform these features and generate a sample graph self-attention transformation vector, which contains important information and dependencies of the input features and is the basis for subsequent predictions.

[0131] After obtaining the sample graph self-attention conversion vector, the server can input it into the prediction unit of the basic business process derivation network. The prediction unit will predict the next software repair application business process that is most likely to jump to under the sample business process path based on the information in the conversion vector, and generate corresponding training derivation data.

[0132] The server can compare the generated training inference data with the previously configured sample business process inference data. By calculating the difference or error between the two, the server can evaluate the prediction performance of the basic business process inference network.

[0133] If the prediction performance does not meet the requirements, the server can adjust the parameters of the network (such as weights, biases, etc.) and repeat the above training process until the prediction performance reaches the preset standard.

[0134] Once the training is completed, the server can output the network parameter information of the trained basic business process derivation network as the network parameter information of the first business process derivation network, so that the first business process derivation network has the ability to derive subsequent business process paths from the business process path data. In subsequent practical applications, the server can use this trained first business process derivation network to perform business process derivation and prediction.

[0135] In a possible implementation, step S101 may include: Step S1011 , merging the second a priori software repair application business process of the a priori critical software repair application business process with the a priori critical software repair application business process to generate first sample business process path data.

[0136] Step S1012, determining a first progress point when terminating the a priori critical software repair application service process and a second progress point when triggering the next software repair application service process of the a priori critical software repair application service process.

[0137] Step S1013: When the difference between the second progress point and the first progress point is greater than the set difference, the sample business process derivation data of the first sample business process path data is output as the first sample business process derivation state; when the difference between the second progress point and the first progress point is not greater than the set difference, the sample business process derivation data of the first sample business process path data is output as the second sample business process derivation state.

[0138] In this embodiment, it is further assumed that the a priori critical software repair application business process is "system upgrade post-performance evaluation." Before this critical process, there are usually one or more related second a priori software repair application business processes, such as "system upgrade deployment" and "new version testing."

[0139] The server may first obtain the historical data of these second prior software repair application business processes, including their execution steps, resource requirements, expected results, etc. Then, the server may merge the "system post-upgrade performance evaluation" with these second prior processes to generate a series of first example business process path data.

[0140] The fusion process may include directly adding certain steps of the second prior process to the initial steps of the "system performance evaluation after upgrade", or readjusting the execution order and dependencies of the steps according to the correlation of historical data. In this way, the generated first-example business process path data will be closer to the actual scenario and help with subsequent network training and derivation.

[0141] When configuring the sample business process to derive data, the server needs to specify two key progress points: the first progress point and the second progress point.

[0142] The first progress point refers to the progress status at the end of the prior critical software repair application business process (such as "performance evaluation after system upgrade"), which can usually be determined by monitoring the execution log, resource usage or output results of the process.

[0143] The second progress point refers to the progress status when the next software repair application business process of the prior critical software repair application business process is triggered. For example, after the "post-system upgrade performance evaluation" is completed, subsequent processes such as "performance tuning" or "troubleshooting" may be triggered. The server needs to determine the specific progress point when these subsequent processes are triggered.

[0144] The server may compare a difference between the second progress point and the first progress point, where the difference reflects a time interval or a logical distance between the prior critical software repair application service process and its subsequent process.

[0145] If the difference is greater than a set difference threshold (this threshold can be set according to actual conditions), the server can consider that there is a certain time interval or logical independence between the two processes. Therefore, the sample business process derivation data of the first sample business process path data can be output as the first sample business process derivation state, which can indicate that the triggering of the subsequent process does not directly depend on the result of the prior key process, or there is a certain uncertainty.

[0146] If the difference is not greater than the set difference threshold, the server may consider that the time interval between the two processes is short or logically closely related. Therefore, the sample business process derivation data of the first sample business process path data may be output as a second sample business process derivation state, which may indicate that the triggering of the subsequent process directly depends on the result of the prior critical process, or has a higher certainty.

[0147] Through this process, the server can configure the appropriate sample business process derivation data state for the first sample business process path data, so that subsequent network training and derivation can more accurately simulate the business process path and derivation relationship in the actual scenario.

[0148] In a possible implementation, step S103 may include: Step S1031: extract a multidimensional feature vector of the first sample business process path feature, and add a position code to each feature vector in the multidimensional feature vector to indicate the position of the feature vector in the first sample business process path feature.

[0149] Step S1032, calculating the attention score between each input feature vector and all other feature vectors in the multi-dimensional feature vector through the graph self-attention unit.

[0150] Step S1033, based on the calculated attention score, a preset mask is applied to shield invalid feature vectors, and a scaling factor is applied to adjust the distribution of the attention score, and then, according to the attention score, each remaining feature vector after the shielding process is weighted summed to generate the sample image self-attention conversion vector.

[0151] In an actual software repair project management scenario, when the server uses the graph self-attention unit of the basic business process inference network to perform graph self-attention conversion on the first sample business process path feature, it will perform a series of detailed steps to generate a sample graph self-attention conversion vector.

[0152] In detail, we continue to assume that the server has been configured with first-instance business process path data related to a priori critical software repair application business processes (such as "performance evaluation after system upgrade"), and first-instance business process path features are extracted from it. These first-instance business process path features may include multiple dimensions such as the execution order of different business steps, resource requirements, and expected results.

[0153] The server can convert these multi-dimensional features into feature vectors, with each dimension corresponding to a feature vector. For example, the execution order dimension may be converted into a vector representing the order of steps, and the resource requirement dimension may be converted into a vector representing the resource type and quantity.

[0154] Since the graph self-attention mechanism cannot directly capture the position information of the feature vector when processing sequence data, the server needs to add a position encoding to each feature vector. Position encoding is a special vector whose value changes according to the position of the feature vector in the sequence and is used to represent the relative or absolute position of the feature vector in the sequence.

[0155] The server can generate a position encoding vector of the same dimension as the feature vector and add or concatenate it with the corresponding feature vector to integrate the position information into the feature vector, so that even if the order of the feature vectors changes during subsequent graph self-attention conversion, the model can recognize their original positions through position encoding.

[0156] The server inputs the multi-dimensional feature vector with position encoding into the graph self-attention unit. The graph self-attention unit calculates the attention score between each input feature vector and all other feature vectors. The calculation of attention scores is usually based on a mechanism called "dot product attention" or "scaled dot product attention". In this mechanism, the graph self-attention unit performs a dot product operation on a feature vector (called the query vector) and all other feature vectors (called the key vector) to obtain a raw attention score. The server can then apply a scaling factor (usually the inverse square root of the feature vector dimension) to adjust the distribution of the raw attention scores to avoid the scores being too large when the dimension is high, causing the softmax function to enter the saturation zone.

[0157] In some cases, some feature vectors may be irrelevant or invalid for the current task. To avoid the influence of these invalid feature vectors on the attention score calculation, the server can apply a preset mask to mask them.

[0158] The preset mask is a binary vector of the same dimension as the feature vector, where the element value corresponding to the invalid feature vector is 0, and the element value corresponding to the valid feature vector is 1. During the attention score calculation process, the server can perform an element-by-element multiplication operation on the mask and the original attention score, thereby setting the attention score of the invalid feature vector to 0.

[0159] After obtaining the scaled and masked attention scores, the server can perform a weighted sum operation on the feature vectors based on these scores to generate a sample graph self-attention transformation vector.

[0160] Specifically, the server can multiply each feature vector (called the value vector) with its corresponding attention score to obtain a weighted feature vector. Then, all weighted feature vectors are summed to obtain the final sample graph self-attention transformation vector, which incorporates the information of all input feature vectors and is weighted according to their attention scores, so as to capture the key information and dependencies in the input sequence.

[0161] Through the above steps, the server can use the graph self-attention unit of the basic business process inference network to perform graph self-attention conversion on the path features of the first sample business process, generate a sample graph self-attention conversion vector containing key information and dependencies, and provide strong support for subsequent business process derivation.

[0162] In a possible implementation, step S130 may include: Step S131: extracting first a priori business process path data from the first a priori software repair application business process.

[0163] Step S132: Generate a second business process path feature according to the first a priori business process path data and the other business process path data.

[0164] Step S133: calling a second business process inference network, where the second business process inference network includes a second graph self-attention unit and a second prediction unit.

[0165] Step S134: Use the second graph self-attention unit to perform graph self-attention conversion on the second business process path feature to generate a second graph self-attention conversion vector.

[0166] Step S135: Utilize the second prediction unit to perform business process derivation on the second graph self-attention conversion vector to generate the second derivation data.

[0167] In this embodiment, in an actual software repair project management scenario, the server can perform business process derivation on the software repair application business process to be triggered to predict the business process that may occur next given the first prior software repair application business process and other potential business process paths.

[0168] In detail, assuming that the software repair application business process to be triggered is "database performance optimization", and the first a priori software repair application business process directly related to it is "database performance monitoring". The server will first extract relevant business process path data from the "database performance monitoring" process, including monitoring time, monitoring indicators, monitoring results, etc., which are called first a priori business process path data.

[0169] In addition to the first a priori business process path data, the server also considers other potential business process path data, which may come from other software repair application business processes related to "database performance optimization", such as "network troubleshooting" or "system security reinforcement".

[0170] The server can fuse the first prior business process path data with other business process path data to generate a comprehensive business process path feature vector, namely, the second business process path feature. The feature vector contains key information extracted from multiple angles and business scenarios, which helps the server to more comprehensively understand the context and possible development direction of the current business process.

[0171] In order to perform business process deduction, the server may call a pre-trained second business process deduction network, where the second business process deduction network is specially designed and trained for other potential business process paths related to the software repair application business process to be triggered.

[0172] The second business process inference network includes a second graph self-attention unit and a second prediction unit. The second graph self-attention unit is used to capture complex dependencies and structural information in the business process path data, and the second prediction unit generates an inference result based on this information.

[0173] The server inputs the second business process path feature into the second graph self-attention unit of the second business process derivation network. The second graph self-attention unit encodes each element in the feature vector and calculates the attention scores between them to measure the relevance and importance between them.

[0174] Based on the calculated attention scores, the second-graph self-attention unit performs a weighted sum operation on the feature vector to generate a new feature representation, namely the second-graph self-attention transformation vector, which captures the key information and dependencies in the business process path data, providing strong support for subsequent predictions.

[0175] Finally, the server inputs the second graph self-attention conversion vector into the second prediction unit of the second business process inference network. The second prediction unit will predict the most likely business process and its confidence level based on the information in the conversion vector and the training data and parameters of the network given the current business process path.

[0176] The prediction unit will output a derivation result, namely the second derivation data, which includes the predicted next business process name, jump probability and related context information. The server can assist in decision making and resource allocation based on this derivation result to ensure the smooth progress of the software repair project.

[0177] In a possible implementation, after step S140, the method further includes: Step S150: outputting the a priori software repair application service process number of the to-be-triggered software repair application service process as a first a priori software repair application service process number.

[0178] Step S160: outputting the a priori software repair application business process number of the a priori critical software repair application business process as a second a priori software repair application business process number.

[0179] Step S170, when the difference between the first a priori software repair application business process number and the second a priori software repair application business process number is less than a preset value, the to-be-triggered software repair application business process is migrated from the target software repair application business process to the other software repair application business process.

[0180] In an actual software repair project management scenario, after the server determines the to-be-triggered software repair application business process as the target software repair application business process, it further considers the historical dependency of the business process to ensure the accuracy of the allocation.

[0181] In detail, assume that the software repair application business process to be triggered is "database performance optimization". After the server determines it as the target software repair application business process (for example, as part of the "post-system upgrade performance evaluation" process), it will review all prior software repair application business processes before "database performance optimization" and calculate the number of these processes, which is recorded as the first prior software repair application business process number.

[0182] For example, if three prior software repair application business processes of "database performance monitoring", "system resource check" and "application log analysis" are performed before "database performance optimization", then the number of the first prior software repair application business process is 3.

[0183] Next, the server can check the a priori software repair application business process number of the a priori critical software repair application business process (in this example, "performance evaluation after system upgrade"). These processes are all related processes performed before "performance evaluation after system upgrade". The server can calculate their number and record it as the second a priori software repair application business process number.

[0184] For example, if three prior software repair application business processes of "system upgrade deployment", "new version testing" and "preliminary performance evaluation" are performed before "system post-upgrade performance evaluation", then the number of the second prior software repair application business process is also 3.

[0185] The server may compare the first a priori software repair application service process number and the second a priori software repair application service process number to check whether the to-be-triggered software repair application service process and the target software repair application service process to which it is assigned are consistent in terms of historical dependency.

[0186] In the above example, both numbers are 3, indicating that "Database Performance Optimization" matches the "Post-System Upgrade Performance Assessment" process to which it is assigned in terms of the number of historical dependencies.

[0187] However, if the difference between the two numbers is less than a preset value (eg, a difference of 1 or 2), the server may consider that the difference may mean that there is a certain mismatch between the to-be-triggered software repair application service process and the target software repair application service process in terms of historical dependency.

[0188] For example, if the first prior software repair application business process number of "Database Performance Optimization" is 2, and the second prior software repair application business process number of "System Post-Upgrade Performance Evaluation" is 4, the difference is 2 (less than the preset value, assuming the preset value is 3), the server may think that although "Database Performance Optimization" is related to "System Post-Upgrade Performance Evaluation", it may not completely belong to this process.

[0189] In this case, the server can migrate "database performance optimization" from the target software repair application business process (that is, the "system post-upgrade performance evaluation" process) to other software repair application business processes. The migration target can be other software repair application business processes that are more compatible with "database performance optimization" in terms of historical dependencies.

[0190] Through this process, the server can more accurately allocate the software repair application business processes to be triggered based on historical dependencies, ensuring the smooth progress of the software repair project.

[0191] Figure 2 The hardware structure of the business process data management system 100 based on artificial intelligence for implementing the business process data management method based on artificial intelligence provided by the embodiment of the present invention is shown as follows: Figure 2 As shown, the artificial intelligence-based business process data management system 100 may include a processor 110 , a machine-readable storage medium 120 , a bus 130 , and a communication unit 140 .

[0192] In an exemplary design concept, the business process data management system 100 based on artificial intelligence can be a single business process data management system based on artificial intelligence, or it can be a server group. The server group can be centralized or distributed (for example, the business process data management system 100 based on artificial intelligence can be a distributed system). In an exemplary design concept, the business process data management system 100 based on artificial intelligence can be local or remote. For example, the business process data management system 100 based on artificial intelligence can access data and / or data stored in the machine-readable storage medium 120 via a network. For another example, the business process data management system 100 based on artificial intelligence can be directly connected to the machine-readable storage medium 120 to access the stored data and / or data. In an exemplary design concept, the business process data management system 100 based on artificial intelligence can be implemented on a cloud platform. As an example only, the cloud platform can include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc. or any combination thereof.

[0193] The machine-readable storage medium 120 can store data and / or instructions. In an exemplary design concept, the machine-readable storage medium 120 can store multidimensional monitoring data obtained from an external terminal. In an exemplary design concept, the machine-readable storage medium 120 can store multidimensional monitoring data and / or instructions that the business process data management system 100 based on artificial intelligence uses to execute or use to complete the exemplary method described in the present invention. In an exemplary design concept, the machine-readable storage medium 120 may include a large-capacity memory, a removable memory, a volatile read-write memory, a read-only memory (ROM), etc. or any combination thereof. Exemplary large-capacity memory may include a disk, an optical disk, a solid-state disk, etc. Exemplary removable memory may include a flash drive, a floppy disk, an optical disk, a memory card, a compressed disk, a tape, etc. Exemplary volatile read-write memory may include a random access memory (RAM). Exemplary RAM may include active random access memory (DRAM), double data rate synchronous active random access memory (DDR SDRAM), passive random access memory (SRAM), thyristor random access memory (T-RAM) and zero capacitance random access memory (Z-RAM), etc. Exemplary read-only memories may include mask read-only memories (MROMs), programmable read-only memories (PROMs), erasable programmable read-only memories (PEROMs), electrically erasable programmable read-only memories (EEPROMs), compact disk read-only memories (CD-ROMs), and digital versatile disk read-only memories, etc. In an exemplary design concept, the machine-readable storage medium 120 may be implemented on a cloud platform. As an example only, a cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc., or any combination thereof.

[0194] During the specific implementation process, at least one processor 110 executes computer executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the business process data management method based on artificial intelligence in the above method embodiment. The processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130, and the processor 110 can be used to control the sending and receiving actions of the communication unit 140.

[0195] The specific implementation process of the processor 110 can refer to the various method embodiments executed by the above-mentioned artificial intelligence-based business process data management system 100. The implementation principles and technical effects are similar, and this embodiment will not be repeated here.

[0196] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer executable instructions are preset. When a processor executes the computer executable instructions, the above-mentioned business process data management method based on artificial intelligence is implemented.

[0197] Similarly, it should be noted that in order to simplify the description of the present invention disclosure and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the invention, sometimes multiple features are grouped into one embodiment, drawings, or descriptions thereof. Similarly, it should be noted that in order to simplify the description of the present invention disclosure and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the invention, sometimes multiple features are grouped into one embodiment, drawings, or descriptions thereof.

Claims

1. A business process data management method based on artificial intelligence, characterized in that: The method comprises: Determine target business process path data of a target software repair application business process based on a priori key software repair application business process of the software repair application business process to be triggered, and determine other business process path data of other software repair application business processes based on a priori other software repair application business processes of the software repair application business process to be triggered, wherein the other software repair application business processes are software repair application business processes other than the target software repair application business process; Performing business process derivation on the first prior software repair application business process of the to-be-triggered software repair application business process and the target business process path data to generate first derivation data, wherein the first derivation data is used to reflect the confidence of jumping to the next software repair application business process after terminating the to-be-triggered software repair application business process when the to-be-triggered software repair application business process is the target software repair application business process; Performing business process derivation on the first a priori software repair application business process of the to-be-triggered software repair application business process and the other business process path data to generate second derivation data, wherein the second derivation data is used to reflect the confidence of jumping to the next software repair application business process after terminating the to-be-triggered software repair application business process when the to-be-triggered software repair application business process is the other software repair application business process; When the deviation confidence between the first derivation data and the second derivation data is greater than the threshold confidence, the software repair application business process to be triggered is output as the target software repair application business process; when the deviation confidence between the first derivation data and the second derivation data is not greater than the threshold confidence, the software repair application business process to be triggered is output as the other software repair application business process.

2. The business process data management method based on artificial intelligence according to claim 1 is characterized in that: The step of determining target business process path data of a target software repair application business process based on a priori key software repair application business process of the software repair application business process to be triggered includes: Acquire a second a priori software repair application service process of the a priori critical software repair application service process, where the second a priori software repair application service process is another software repair application service process before the a priori critical software repair application service process; Performing regular conversion on the business process connection degree between the second a priori software repair application business process and the first a priori software repair application business process to generate a business process influence coefficient of the a priori key software repair application business process; The prior critical business process path data is extracted from the prior critical software repair application business process, and the target business process path data of the target software repair application business process is determined based on the business process impact coefficient of the prior critical software repair application business process and the prior critical business process path data.

3. The business process data management method based on artificial intelligence according to claim 2 is characterized in that: Before determining the target business process path data of the target software repair application business process based on the business process impact coefficient of the a priori critical software repair application business process and the a priori critical business process path data, the method further includes: Acquire a target business process participating node corresponding to the software repair application business process to be triggered, and acquire a reference business process participating node having a business process connection degree with the target business process participating node greater than a preset connection degree from a business process participating node sequence, wherein the business process participating node sequence includes a plurality of business process participating nodes; Extracting reference a priori critical software repair application business process path data from the reference business process participating node's a priori critical software repair application business process, and determining a business process impact coefficient of the reference business process participating node's a priori critical software repair application business process; The determining, based on the business process impact coefficient of the a priori critical software repair application business process and the a priori critical business process path data, the target business process path data of the target software repair application business process comprises: According to the business process influence coefficient of the a priori key software repair application business process of the software repair application business process to be triggered, the a priori key business process path data of the software repair application business process to be triggered are merged to generate the first business process path data of the target software repair application business process; According to the business process influence coefficient of the a priori critical software repair application business process of the reference business process participating node, the a priori critical business process path data of the reference business process participating node are integrated to generate the second business process path data of the target software repair application business process; Integrating the first business process path data and the second business process path data to generate target business process path data of the target software repair application business process; The step of obtaining, from the sequence of business process participating nodes, a reference business process participating node having a business process connection degree with the target business process participating node greater than a preset connection degree comprises: Acquire a database of a sequence of nodes participating in a business process, wherein the database records information of all nodes that have participated in the software repair application business process, and the node information includes an identifier of the node participating in the business process, the software repair application business process that it has participated in historically, and performance and evaluation data of the software repair application business process; For each candidate business process participating node in the business process participating node sequence, analyzing the candidate software repair application business processes that the business process participating node has historically participated in and that have an intersection with the target business process participating node; Obtaining the number of software repair application business processes and the similarity of participation types in which each candidate business process participating node and the target business process participating node jointly participate in the software repair application business process data of the candidate software repair application business process, and obtaining first evaluation data; Obtaining the sequence and dependency relationship between each candidate business process participating node and the target business process participating node in the jointly participating candidate software repair application business process to obtain second evaluation data; Obtaining the time proximity of the candidate software repair application business process jointly participated by each candidate business process participating node and the target business process participating node to obtain third evaluation data; Obtaining the similarity of performance evaluation between each candidate business process participating node and the target business process participating node in the jointly participating candidate software repair application business process, to obtain fourth evaluation data; According to the first evaluation data, the second evaluation data, the third evaluation data and the fourth evaluation data, corresponding evaluation weights are set and weighted fusion is performed to generate a business process connection degree between each candidate business process participating node and a target business process participating node; The business process connection degree is compared with a preset connection degree, and when the business process connection degree is greater than the preset connection degree, the corresponding candidate business process participating node is output as the reference business process participating node.

4. The business process data management method based on artificial intelligence according to claim 2 is characterized in that: Before performing regular conversion on the business process connection degree between the second a priori software repair application business process and the first a priori software repair application business process to generate the business process influence coefficient of the a priori key software repair application business process, the method further includes: For each of the second a priori software repair application service processes, obtaining second a priori service process path data from the second a priori software repair application service process; Obtaining first a priori business process path data from the first a priori software repair application business process of the to-be-triggered software repair application business process; The business process connection degree between the first a priori business process path data and the second a priori business process path data is output as the business process connection degree between the second a priori software repair application business process and the first a priori software repair application business process.

5. The business process data management method based on artificial intelligence according to claim 1 is characterized in that: Before determining other service process path data of other software repair application service processes based on the prior other software repair application service processes of the to-be-triggered software repair application service process, the method further includes: Decomposing the priori software repair application service process of the to-be-triggered software repair application service process into time periods to generate a plurality of priori software repair application service process time periods, each of the priori software repair application service process time periods including at least one of the priori software repair application service processes; When the final software repair application service process in the priori software repair application service process period is the other software repair application service process, the final software repair application service process in the priori software repair application service process period is output as the priori other software repair application service process of the software repair application service process to be triggered.

6. The business process data management method based on artificial intelligence according to claim 1 is characterized in that: The performing business process derivation on the first a priori software repair application business process of the to-be-triggered software repair application business process and the target business process path data to generate first derivation data includes: Extracting first a priori business process path data from the first a priori software repair application business process; Determine a first business process path feature according to the first priori business process path data and the target business process path data; Calling a first business process inference network, wherein the first business process inference network includes a first graph self-attention unit and a first prediction unit; Using the first graph self-attention unit to perform graph self-attention conversion on the first business process path feature to generate a first graph self-attention conversion vector; Using the first prediction unit to perform business process derivation on the first graph self-attention conversion vector to generate the first derivation data; The determining of the first business process path feature according to the first priori business process path data and the target business process path data includes: Fusion of the first priori business process path data and the target business process path data to generate the first business process path feature; Performing weighted processing on the first priori business process path data and the target business process path data to generate the first business process path feature; The target business process participating node vector corresponding to the to-be-triggered software repair application business process is obtained, and the first priori business process path data, the target business process path data and the target business process participating node vector are integrated to generate the first business process path feature.

7. The business process data management method based on artificial intelligence according to claim 6 is characterized in that: Before calling the first service process to derive the network, the method further includes: configuring first sample business process path data and sample business process derivation data of the first sample business process path data according to the a priori critical software repair application business process; Extracting first sample business process path features from the first sample business process path data; Using the graph self-attention unit of the basic business process derivation network, perform graph self-attention conversion on the first sample business process path feature to generate a sample graph self-attention conversion vector; Using the prediction unit of the basic business process derivation network, the business process is deduced for the sample graph self-attention conversion vector to generate training derivation data; Training the basic business process derivation network according to the training derivation data and the sample business process derivation data, and outputting the network parameter information of the trained basic business process derivation network as the network parameter information of the first business process derivation network; The configuring of the first sample business process path data and the sample business process derivation data of the first sample business process path data according to the prior critical software repair application business process includes: Merging a second a priori software repair application business process of the a priori critical software repair application business process with the a priori critical software repair application business process to generate first sample business process path data; Determine a first progress point when the a priori critical software repair application service process is terminated and a second progress point when the next software repair application service process of the a priori critical software repair application service process is triggered; When the difference between the second progress point and the first progress point is greater than a set difference, the sample business process derivation data of the first sample business process path data is output as a first sample business process derivation state; when the difference between the second progress point and the first progress point is not greater than the set difference, the sample business process derivation data of the first sample business process path data is output as a second sample business process derivation state; The step of using the graph self-attention unit of the basic business process derivation network to perform graph self-attention conversion on the first sample business process path feature to generate a sample graph self-attention conversion vector includes: Extracting a multidimensional feature vector of the first sample business process path feature, and adding a position code to each feature vector in the multidimensional feature vector to indicate the position of the feature vector in the first sample business process path feature; Calculating the attention score between each input feature vector and all other feature vectors in the multi-dimensional feature vector by the graph self-attention unit; Based on the calculated attention score, a preset mask is applied to shield invalid feature vectors, and a scaling factor is applied to adjust the distribution of the attention score. Then, according to the attention score, each remaining feature vector after the shielding process is weighted summed to generate the sample image self-attention conversion vector.

8. The business process data management method based on artificial intelligence according to claim 1 is characterized in that: The performing business process derivation on the first a priori software repair application business process of the to-be-triggered software repair application business process and the other business process path data to generate second derivation data includes: Extracting first a priori business process path data from the first a priori software repair application business process; generating a second business process path feature according to the first a priori business process path data and the other business process path data; Calling a second business process inference network, wherein the second business process inference network includes a second graph self-attention unit and a second prediction unit; Using the second graph self-attention unit to perform graph self-attention conversion on the second business process path feature to generate a second graph self-attention conversion vector; The second prediction unit is used to derive the business process of the second graph self-attention conversion vector to generate the second derivation data.

9. The business process data management method based on artificial intelligence according to claim 1 is characterized in that: After outputting the to-be-triggered software repair application service process as the target software repair application service process, the method further includes: Outputting the a priori software repair application service process number of the to-be-triggered software repair application service process as a first a priori software repair application service process number; Outputting the a priori software repair application business process number of the a priori critical software repair application business process as a second a priori software repair application business process number; When the difference between the first a priori software repair application business process number and the second a priori software repair application business process number is less than a preset value, the to-be-triggered software repair application business process is migrated from the target software repair application business process to the other software repair application business process.

10. A business process data management system based on artificial intelligence, characterized in that: The artificial intelligence-based business process data management system includes a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions, and the machine-executable instructions are loaded and executed by the processor to implement the artificial intelligence-based business process data management method of any one of claims 1-9.

Images