Alignment-based hierarchical business process model compliance detection method
By adopting an alignment-based hierarchical business process model compliance detection method, the efficiency problem of hierarchical structure model detection is solved, achieving efficient compliance detection and reducing time and memory consumption.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-26
- Publication Date
- 2026-04-03
AI Technical Summary
Existing process model compliance detection technologies struggle to efficiently detect hierarchical business process models, especially when matching large-scale log files with the models, consuming significant amounts of time and memory.
An alignment-based hierarchical business process model compliance detection method is adopted. This method reduces the alignment time between logs and the model by mining hierarchical transition nesting relationships, constructing hierarchical event logs, constructing an alignment sequence relationship nesting tree, and merging alignment sequences.
It effectively reduces the time overhead of compliance testing for layered models, improves the efficiency of model repair and optimization, and reduces the time consumed in alignment.
Smart Images
Figure CN115904748B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of process mining, and more specifically to a compliance detection method for a hierarchical business process model based on alignment. Background Technology
[0002] Process mining, a new research hotspot in the field of Business Process Management, aims to bridge the gap between traditional model-driven methods (such as business process modeling and model correctness verification) and novel data-driven methods (such as data mining and machine learning). Compliance testing is a crucial component of process mining, a technique used to assess the degree of matching between process models and logs—that is, the extent to which the model can replay the essence of the log-related process execution. Typically, enterprises do not implement systematic modeling in the initial stages, but their internal software systems still record a large number of business activity execution processes. As enterprises grow in size and their business becomes more complex, they adopt automated mining techniques to extract work process models from a large amount of actual logs to better manage performance and improve efficiency. Compliance checks are usually required to evaluate the quality of the mined models. Furthermore, when the logs generated by the process model deviate significantly from the original model—that is, when the fit is below a given threshold—it is necessary to correct the original model.
[0003] Existing process model compliance detection techniques: (1) Dr. Arya Adriansyah proposed constructing a product net of the process model and the corresponding process model of the trajectory, and then constructing a reachability graph of the product net. The problem of optimal alignment is transformed into the problem of searching for the optimal path from the initial state to the final state in the reachability graph, which can be solved by the A* search algorithm. The fit evaluation index is used to evaluate the quality of the process model. (2) de Leoni et al. proposed a method for aligning event logs and declarative models, and provided a complex diagnostic method that accurately points out the location and severity of deviations. The above alignment methods are all defined on the basis of flat models and are not applicable to hierarchical models with hierarchical structures. Dr. Joos Buijs discussed in detail the relationship between the four metrics of fit, accuracy, generality and complexity and their importance for model quality evaluation. These compliance evaluation indicators are all defined on the basis of flat process models and are not applicable to hierarchical models containing sub-process behaviors.
[0004] To address this issue, Associate Professor Wen Lijie and colleagues at Tsinghua University proposed Acorn, a compliance detection algorithm based on the BPMN model, which is solely based on the BPMN model. Professor Liu Cong of Shandong University of Technology proposed a method to transform a hierarchical model with subprocesses into a flat model. However, in the real world, log files and their corresponding models are usually quite large, making this method of transforming a hierarchical model into a flat model for compliance detection consume significant time and memory. Summary of the Invention
[0005] To address the aforementioned problems in existing technologies, this invention provides a compliance detection method for a hierarchical business process model based on alignment.
[0006] The present invention adopts the following technical solution:
[0007] The alignment-based hierarchical business process model compliance detection method includes the following steps:
[0008] Step 1: Hierarchical transition nesting relationship mining; The hierarchical process model hpn is used as input to mine the nesting relationship between models. The output is a hierarchical transition nesting relationship tree ang, which describes all possible nesting relationships in the hierarchical model.
[0009] Step 2: Construction of hierarchical event log; Taking the event log xlog with lifecycle and the hierarchical change nesting relationship tree ang as input, the hierarchical event log hlog is constructed by analyzing the hierarchical nesting relationship of the model;
[0010] Step 3: Construct an aligned sequence nested tree; using the hierarchical event log hlog and the hierarchical process model hpn as input, perform compliance checks on the hierarchical structure to obtain the aligned sequence nested tree hat;
[0011] Step 4: Merge the hierarchical alignment sequences; merge the nested relationship trees of the alignment sequences obtained in Step 3 to obtain the final result.
[0012] Preferably, step 1 specifically includes:
[0013] Input: Hierarchical process model hpn;
[0014] Output: Hierarchical transition nested relation tree (ang);
[0015] Step 1.1: Call the algorithms getActivityNestedSet() and getActivityPair() to mine the nested transition set activityNestedSet[] and its nested transition pair set activityPariSet[] in the hierarchical model;
[0016] Step 1.2: By calling the algorithm ActivityGraphConstruction(), construct a hierarchical nested transition relationship tree ang based on the nested transition associations;
[0017] Step 1.3: Return the hierarchical transition nested relationship tree ang.
[0018] Preferably, the getActivityNestedSet() function is called to return all nested transitions in the hierarchical model, specifically described as follows:
[0019] Input: Hierarchical process model hpn;
[0020] Output: a nested transition collection activityNestedSet[];
[0021] Traverse the top-level Petri net, find the nested transitions it contains and store them in the collection activitySet[]. Recursively traverse the nested layered Petri nets of the top-level Petri net and return the collection of all nested transitions in the layered model.
[0022] The getActivityPair() function is called to return nested transition pairs in a hierarchical model, specifically described as follows:
[0023] Input: Hierarchical process model hpn;
[0024] Output: A collection of nested transition pairs, activityPairSet[];
[0025] Extract nested transitions t from the hierarchical nested transition relationship tree ang, assign nested transition t to the variable source, and find the hierarchical model hpn nested within nested transition t. i Find the hierarchical model hpn i Nested transitions t i Store the (source, target) nested transition association pairs in the activityPairSet[].
[0026] Recursive traversal of the nested transitions of the top-level Petri net, nested layered Petri net hierarchical model hpn i Returns a collection called activityPairSet[];
[0027] The called activityGraphConstruction() function returns the nested transition tree ang in the hierarchical model, specifically described as follows:
[0028] Input: A collection of nested transition associative pairs, activityPariSet[];
[0029] Output: Hierarchical transition nested relation tree (ang);
[0030] Extract nested transition pairs sequentially from activityPairSet[], store the nodes and corresponding edges in the nested transition pairs into a tree structure; return the hierarchical nested transition relationship tree ang.
[0031] Preferably, step 2 specifically includes:
[0032] Input: a hierarchical change nested relation tree (ang), and an event log (xlog) with lifecycle;
[0033] Output: Hierarchical event log hlog;
[0034] Extract nodes sequentially from the hierarchical transition nested relationship tree ang and store them in allNestedActivities[];
[0035] Based on the nodes in the hierarchical transition nested relation tree ang, the root log activity and the top-level nested log activities under the root log activity are established respectively.
[0036] Extract the log activity from the lifecycle-bound event log xlog corresponding to the root log activity and assign it to mainLog;
[0037] The log activity corresponding to the top-level nested log activity under the root log activity is extracted from the event log xlog with a lifecycle and assigned to submainLog;
[0038] Recursively traverse other nodes in the hierarchical transition nested relation tree ang;
[0039] Returns the hierarchical event log hlog.
[0040] Preferably, step 3 specifically includes:
[0041] Input: Hierarchical process model (hpn), hierarchical event log (hlog);
[0042] Output: Aligned sequence nested relationship tree (hat);
[0043] Extract the root log (roothlog) from the hierarchical event log (hlog);
[0044] Extract the top-level Petir net pn from the hierarchical model;
[0045] Align pn with roothlog to obtain the top-level aligned sequence rootA;
[0046] A hierarchical model of nested top-level Petri nets (hpn) iThe hierarchical event log roothlog nested within the root log i Recursively calculate the alignment sequence and return the alignment sequence nested relationship tree (hat).
[0047] Preferably, step 4 specifically includes:
[0048] Input: an aligned sequence nested relation tree (hat);
[0049] Output: Aligned sequence r;
[0050] When the alignment sequence nesting tree hat is not empty, extract the top-level alignment sequence from the alignment sequence nesting tree hat and push it into the set rootA;
[0051] Push the elements in set rootA into queue st;
[0052] Extract each element from queue st sequentially and check if it is a nested transition, i.e., tag. i Is it 1?
[0053] If tag i If the value is not 1, push the current node into queue r;
[0054] If tag i =1, push the current node into queue r, and extract the aligned sequence nested relation tree hat. i Align the top-level sequence and push it into the set rootA; recursively traverse hat i
[0055] Until queue
[0056] Returns the final aligned sequence r.
[0057] The beneficial effects of this invention are:
[0058] The alignment-based hierarchical business process model compliance detection method proposed in this invention is a novel approach for hierarchical model compliance detection. Compared to existing hierarchical model compliance detection methods, this method reduces the alignment time between logs and the model, which helps to reduce the time overhead required for model repair and optimization. The alignment-based hierarchical business process model compliance detection method requires less time than existing hierarchical model compliance detection techniques. Attached Figure Description
[0059] Figure 1 This is an example of a hierarchical process model hpn in Example 1.
[0060] Figure 2 According to Figure 1 The nested relationship tree of hierarchical transitions obtained from the hierarchical process model.
[0061] Figure 3 The hierarchical process model hpn1 is shown in Example 1.
[0062] Figure 4 For hierarchical event logs, hl1.
[0063] Figure 5 The hierarchical process model is hpn2.
[0064] Figure 6 For hierarchical event logs, use hl2.
[0065] Figure 7 The alignment sequence nesting relationship tree hat obtained in Example 1.
[0066] Figure 8 This is a schematic diagram of merging and aligning sequences.
[0067] Figure 9 A schematic diagram for integrating r1 aligned sequences.
[0068] Figure 10 A schematic diagram for integrating r2-aligned sequences.
[0069] Figure 11 A schematic diagram for integrating r3 aligned sequences.
[0070] Figure 12 The time cost is when the noise threshold is 2.0.
[0071] Figure 13 The time cost is when the noise threshold is 3.0.
[0072] Figure 14 The time cost is when the noise threshold is 5.0.
[0073] Figure 15 This incurs time overhead for different noise thresholds when the log size is fixed. Detailed Implementation
[0074] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings and specific examples:
[0075] Combination Figures 1 to 11 The compliance detection method based on the alignment-based hierarchical business process model includes the following steps:
[0076] Step 1: Hierarchical transition nesting relationship mining; The hierarchical process model hpn is used as input to mine the nesting relationship between models. The output is a hierarchical transition nesting relationship tree ang, which describes all possible nesting relationships in the hierarchical model.
[0077] Specifically, it includes:
[0078] Input: Hierarchical process model hpn;
[0079] Output: Hierarchical transition nested relation tree (ang);
[0080] Step 1.1: Call the algorithms getActivityNestedSet() and getActivityPair() to mine the nested transition set activityNestedSet[] and its nested transition pair set activityPariSet[] in the hierarchical model;
[0081] Step 1.2: By calling the algorithm ActivityGraphConstruction(), construct a hierarchical nested transition relationship tree ang based on the nested transition associations;
[0082] Step 1.3: Return the hierarchical transition nested relationship tree ang.
[0083] The getActivityNestedSet() function, which is called, primarily returns all nested transitions in the hierarchical model. Specifically, it is described as follows:
[0084] Input: Hierarchical process model hpn;
[0085] Output: a nested transition collection activityNestedSet[];
[0086] Traverse the top-level Petri net, find its nested transitions and store them in the collection activitySet[]. Recursively traverse the nested layered Petri nets of the top-level Petri net and return the collection of all nested transitions in the layered model.
[0087] The getActivityPair() function, when called, primarily returns nested transition pairs in the hierarchical model, specifically described as follows:
[0088] Input: Hierarchical process model hpn;
[0089] Output: A collection of nested transition pairs, activityPairSet[];
[0090] Extract nested transitions t from the hierarchical nested transition relationship tree ang, assign nested transition t to the variable source, and find the hierarchical model hpn nested within nested transition t. i Find the hierarchical model hpn i Nested transitions t i Store it in target; store the nested transition pair (source, target) in the collection activityPairSet[].
[0091] Recursive traversal of the nested transitions of the top-level Petri net, nested layered Petri net hierarchical model hpn i Returns a collection activityPairSet[].
[0092] The called activityGraphConstruction() function primarily returns the nested transition tree ang in the hierarchical model, specifically described as follows:
[0093] Input: A collection of nested transition associative pairs, activityPariSet[];
[0094] Output: Hierarchical transition nested relation tree (ang);
[0095] Extract nested transition pairs sequentially from activityPairSet[], store the nodes and corresponding edges in the nested transition pairs into a tree structure; return the hierarchical nested transition relationship tree ang.
[0096] for Figure 1 An example of a hierarchical process model hpn is given, and its hierarchical transition nested relationship tree is obtained as follows: Figure 2 As shown, (1) a and c are root nodes; (2) a and b satisfy a nested relationship, and b and e satisfy a nested relationship.
[0097] Step 2: Constructing a hierarchical event log; Taking the event log xlog with a lifecycle and the hierarchical transition nested relationship tree ang as input, the hierarchical event log hlog is constructed by analyzing the hierarchical nested relationship of the model.
[0098] Specifically, it includes:
[0099] Input: a hierarchical change nested relation tree (ang), and an event log (xlog) with lifecycle;
[0100] Output: Hierarchical event log hlog;
[0101] Extract nodes sequentially from the hierarchical transition nested relationship tree ang and store them in allNestedActivities[];
[0102] Based on the nodes in the hierarchical transition nested relation tree ang, the root log activity and the top-level nested log activities under the root log activity are established respectively.
[0103] Extract the log activity from the lifecycle-bound event log xlog corresponding to the root log activity and assign it to mainLog;
[0104] The log activity corresponding to the top-level nested log activity under the root log activity is extracted from the event log xlog with a lifecycle and assigned to submainLog;
[0105] Recursively traverse other nodes in the hierarchical transition nested relation tree ang;
[0106] Returns the hierarchical event log hlog.
[0107] by Figure 3 The hierarchical process model hpn1 and its lifecycle-bound event log L1 = { s ,b s ,b c ,a c > 90 , s ,b s ,a c ,b c > 1 For example, the hierarchical model hpn1 and its lifecycle-bound event log L1 = { s ,b s ,b c ,a c > 90 , s ,b s ,a c ,b c > 1} is the input, which can be used to generate a layered event log hl1, such as Figure 4 As shown, its root log is rootLog = { s ,a c > 91}, its nested task set NA(rootLog) = {a}. The sub-log corresponding to nested task a is NLoga = { s ,b c > 90}
[0108] Step 3: Construct an aligned sequence nested tree; using the hierarchical event log hlog and the hierarchical process model hpn as input, perform compliance checks on the hierarchical structure to obtain the aligned sequence nested tree hat;
[0109] Input: Hierarchical process model (hpn), hierarchical event log (hlog);
[0110] Output: Aligned sequence nested relationship tree (hat);
[0111] Extract the root log (roothlog) from the hierarchical event log (hlog);
[0112] Extract the top-level Petir net pn from the hierarchical model;
[0113] Align pn with roothlog to obtain the top-level aligned sequence rootA;
[0114] A hierarchical model of nested top-level Petri nets (hpn) i The hierarchical event log roothlog nested within the root log i Recursively calculate the alignment sequence and return the alignment sequence nested relationship tree (hat).
[0115] by Figure 5 The hierarchical process model hpn2 and its lifecycle-bound event log L2 = { s ,b s ,d s ,d c ,b c ,a c ,c s ,c c > 99 , <c s ,a s ,b s ,c c ,d s ,d c ,b c ,a c > 96 , s ,c s ,b s ,d s ,d c ,b c ,a c ,c c > 86 , s ,b s ,d s ,c s ,d c ,b c ,a c ,c c > 78 , s ,b s ,d s ,d c ,c s ,b c ,a c ,c c > 79 , s ,b s ,d s ,d c ,b c ,cs ,a c ,c c > 82 , s ,c s ,b s ,d s ,c c ,d c ,b c ,a c > 98 , s ,b s ,c c ,d s ,d c ,b c ,c c ,a c > 85 , s ,b s ,c s ,d s ,c c ,d c ,b c ,a c ,c c > 100 For example,}
[0116] Taking the hierarchical model hpn2 and the lifecycle-bound log L2 as input, the hierarchical event log hl2 is obtained. Its root log is rootLog = { s ,a c ,c s ,c c > 99 , <c s ,a s ,c c ,a c > 96 , s ,c s ,a c ,c c > 86 , s ,c s ,a c ,c c > 78 , s ,c s ,a c ,c c > 79 , s ,c s ,a c ,c c >82 , s ,c s ,c c ,a c > 98 , s ,c c ,c c ,a c > 85 , s ,c s ,c c ,a c ,c c > 100 The nested task set NA(rootLog) = {a} is given by the given task set NA(rootLog). The sub-log corresponding to nested task a is NLog. a ={ s ,b c > 99 , s ,b c > 96 , s ,b c > 86 , s ,b c > 78 , s ,b c > 79 , s ,b c > 82 , s ,b c > 98 , s ,b c > 85 , s ,b c > 100}, whose nested task set NA(NLog a = {b}. The sub-log NLog corresponding to nested task b. b ={ <d s ,d c > 99 , <d s ,d c > 96 , <d s ,d c > 86 , <d s ,d c > 78 , <d s ,d c > 79 , <d s ,d c > 82 , <d s ,d c > 98 , <d s ,d c > 85 , <d s ,d c > 100 The nested relationships they have are as follows: Figure 6 As shown.
[0117] Using the hierarchical process model hpn2 and the hierarchical event log hl2 as input, we obtain the aligned sequence nested relation tree hat, as follows: Figure 7 As shown.
[0118] Step 4: Merge the hierarchical alignment sequences; merge the nested relationship trees of the alignment sequences obtained in Step 3 to obtain the final result.
[0119] Specifically, it includes:
[0120] Input: an aligned sequence nested relation tree (hat);
[0121] Output: Aligned sequence r;
[0122] When the alignment sequence nesting tree hat is not empty, extract the top-level alignment sequence from the alignment sequence nesting tree hat and push it into the set rootA;
[0123] Push the elements in set rootA into queue st;
[0124] Extract each element from queue st sequentially and check if it is a nested transition, i.e., tag. i Is it 1?
[0125] If tag i If the value is not 1, push the current node into queue r;
[0126] If tag i =1, push the current node into queue r, and extract the aligned sequence nested relation tree hat. i Align the top-level sequence and push it into the set rootA; recursively traverse hat i
[0127] Until queue
[0128] Returns the final aligned sequence r.
[0129] by Figure 7Taking the nested alignment sequence tree `hat` as an example, we integrate its hierarchical alignment sequences. Elements in `rootA = {r1, r2, r3, ...}` are sequentially pushed into queue `st`; the head element `r1` is popped from queue `st`, and the head pointer points to the successor node `r` of the head element. We then traverse `r1` and the tag of node `a`. i == 1, meaning node a is a nested transition. Push the elements before node a into the queue, recursively traverse the nested hierarchical alignment sequence of node a, and similarly traverse the other nodes in rootA. The specific steps are as follows: Figure 8 , Figure 9 , Figure 10 , Figure 11 As shown.
[0130] Example 1
[0131] Combination Figures 12 to 15 The input data for the hierarchical business process mining method consists of event logs that record lifecycle and sub-process information. This embodiment selects two simulated hierarchical business process models and two real hierarchical business process models as the evaluation data for this invention. The following is a detailed description of the hierarchical business process models:
[0132] The two simulation models are hpn1 and hpn2 mentioned above. The two datasets corresponding to the real hierarchical business process model are the publicly available TSEC Log and CRMC Log.
[0133] Source of the dataset:
[0134] (1) TSEC Log: This dataset is generated based on cross-border e-commerce scenarios, and the process involves two sub-processes.
[0135] (2) CRMC Log: This dataset was generated based on the upgrade process of NetflixAsgard, an open-source cloud resource management tool on Amazon web services, which involves a sub-process.
[0136] The basic information of the dataset is shown in Table 1.
[0137] Table 1. Basic information about the dataset
[0138] Log name Total number of trajectories Total number of events Number of activities <![CDATA[L1]]> 101 214 2 <![CDATA[L2]]> 803 3223 4 TSEC Log 522 14616 10 CRMC Log 626 27544 17
[0139] Evaluation indicators, experimental results and analysis
[0140] Log size assessment
[0141] To test the effectiveness and usability of the compliance detection technology proposed in this paper, logs of different sizes and with different noise thresholds were used as input and compared with existing methods (Convert a Hierarchical Petri Net to a FlatPetri Net). The performance of the algorithm was demonstrated by comparing the time overhead.
[0142] like Figure 12 Experiments show that, with a noise threshold of 2.0, the time overhead of the alignment-based hierarchical business process model compliance detection method increases as the log size grows. The time overhead of L1 with smaller log size is significantly less than that of CRMC Log with larger log size. By comparison, it is clear that the average time of the method proposed in this invention is less than the average time of the previously proposed Convert a Hierarchical Petri Net to a Flat Petri Net algorithm.
[0143] like Figure 13 Experiments show that, with a noise threshold of 3.0, the time overhead of the alignment-based hierarchical business process model compliance detection method increases with the log size. By comparison, the average time of the method proposed in this invention is less than that of the previously proposed algorithm "Convert a Hierarchical Petri Net to a Flat Petri Net".
[0144] like Figure 14 Experiments show that, using the alignment-based hierarchical business process model compliance detection method, with a noise threshold of 5.0, the time overhead increases as the log size increases. The average time of the method proposed in this invention is less than that of the previously proposed algorithm for converting a hierarchical Petri Net to a flat Petri Net.
[0145] according to Figure 12 , Figure 13 , Figure 14 The experimental comparison results show that, ignoring other factors and considering only alignment, under the same noise threshold, the larger the log size, the longer the processing time. The method proposed in this invention has an average processing time that is less than the previously proposed algorithm for converting a hierarchical Petri Net to a flat Petri Net.
[0146] according to Figure 15The experimental results and comparative analysis lead to the following conclusions: Ignoring other cases and considering only alignment, under the same log size, the larger the noise threshold, the longer the processing time. The average processing time of the algorithm proposed in this paper is less than that of the previously proposed "Convert a Hierarchical Petri Net to a Flat Petri Net" algorithm.
[0147] The above experiments show that the alignment-based hierarchical business process model compliance detection technology proposed in this paper requires less time than existing hierarchical model compliance detection technologies, thus verifying the effectiveness and usability of the alignment-based hierarchical business process model compliance detection technology.
[0148] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.
Claims
1. A compliance detection method for a hierarchical business process model based on alignment, characterized in that, Includes the following steps: Step 1: Mining nested relationships in hierarchical transitions; The hierarchical process model hpn is used as input to mine the nesting relationships between models. The output is a hierarchical transition nesting relationship tree ang, which describes all possible nesting relationships in the hierarchical model. Specifically, it includes: Input: Hierarchical process model hpn; Output: Hierarchical transition nested relation tree (ang); Step 1.1: Call the algorithms getActivityNestedSet() and getActivityPair() to mine the nested transition set activityNestedSet[] and its nested transition pair set activityPariSet[] in the hierarchical model; Step 1.2: By calling the algorithm ActivityGraphConstruction(), construct a hierarchical nested transition relationship tree ang based on the nested transition associations; Step 1.3: Return the hierarchical transition nested relationship tree (ang); The `getActivityNestedSet()` function is called to return all nested transitions in the hierarchical model, specifically described as follows: Input: Hierarchical process model hpn; Output: a nested transition collection activityNestedSet[]; Traverse the top-level Petri net, find the nested transitions it contains and store them in the collection activitySet[]. Recursively traverse the nested layered Petri nets of the top-level Petri net and return the collection of all nested transitions in the layered model. The getActivityPair() function is called to return nested transition pairs in a hierarchical model, specifically described as follows: Input: Hierarchical process model hpn; Output: A collection of nested transition pairs, activityPairSet[]; Extract nested transitions t from the hierarchical nested transition relationship tree ang, assign nested transition t to the variable source, and find the hierarchical model hpn nested within nested transition t. i Find the hierarchical model hpn i Nested transitions t i And store it in target; store the nested transition association pair (source, target) in the collection activityPairSet[]; Recursive traversal of the nested transitions of the top-level Petri net, nested layered Petri net hierarchical model hpn i Returns a collection called activityPairSet[]; The called activityGraphConstruction() function returns the nested transition tree ang in the hierarchical model, specifically described as follows: Input: A collection of nested transition associative pairs, activityPariSet[]; Output: Hierarchical transition nested relation tree (ang); Extract nested transition pairs sequentially from activityPairSet[], store the nodes and corresponding edges in the nested transition pairs into a tree structure; return the hierarchical nested transition relationship tree ang; Step 2: Construction of hierarchical event log; Taking the event log xlog with lifecycle and the hierarchical change nesting relationship tree ang as input, the hierarchical event log hlog is constructed by analyzing the hierarchical nesting relationship of the model; Step 3: Construct an aligned sequence nested tree; using the hierarchical event log hlog and the hierarchical process model hpn as input, perform compliance checks on the hierarchical structure to obtain the aligned sequence nested tree hat; Specifically, it includes: Input: Hierarchical process model (hpn), hierarchical event log (hlog); Output: Aligned sequence nested relationship tree (hat); Extract the root log (roothlog) from the hierarchical event log (hlog); Extract the top-level Petir net pn from the hierarchical model; Align pn with roothlog to obtain the top-level aligned sequence rootA; A hierarchical model of nested top-level Petri nets (hpn) i The hierarchical event log roothlog nested within the root log i Recursively calculate the alignment sequence and return the alignment sequence nesting relationship tree (hat). Step 4: Merge the hierarchical alignment sequences; merge the nested relationship trees of the alignment sequences obtained in Step 3 to obtain the final result.
2. The compliance detection method for a hierarchical business process model based on alignment according to claim 1, characterized in that, Step 2 specifically includes: Input: a hierarchical change nested relation tree (ang), and an event log (xlog) with lifecycle; Output: Hierarchical event log hlog; Extract nodes sequentially from the hierarchical transition nested relationship tree ang and store them in allNestedActivities[]; Based on the nodes in the hierarchical transition nested relation tree ang, the root log activity and the top-level nested log activities under the root log activity are established respectively. Extract the log activity from the lifecycle-bound event log xlog corresponding to the root log activity and assign it to mainLog; The log activity corresponding to the top-level nested log activity under the root log activity is extracted from the event log xlog with a lifecycle and assigned to submainLog; Recursively traverse other nodes in the hierarchical transition nested relation tree ang; Returns the hierarchical event log hlog.
3. The compliance detection method for a hierarchical business process model based on alignment according to claim 1, characterized in that, Step 4 specifically includes: Input: an aligned sequence nested relation tree (hat); Output: Aligned sequence r; When the alignment sequence nesting tree hat is not empty, extract the top-level alignment sequence from the alignment sequence nesting tree hat and push it into the set rootA; Push the elements in set rootA into queue st; Extract each element from queue st sequentially and check if it is a nested transition, i.e., tag. i Is it 1? If tag i ≠ 1, push the current node into queue r; If tag i = 1, push the current node into queue r, and extract the aligned sequence nested relation tree hat. i Align the top-level sequence and push it into the set rootA; recursively traverse hat i Until queue st = Returns the final aligned sequence r.