Multi-dimensional process variant difference analysis method and system based on event log

Abstract

The present invention discloses a multi-dimensional process variant difference analysis method and system based on event logs, including: 1) obtaining standardized event logs L1 and L2 recording track, activity, event, resource and timestamp information; 2) using event log L1 as input, converting it into a direct follow-up graph M1 using a flowchart algorithm; 3) adding event log L2 for difference comparison with event log L1, and converting it into a direct follow-up graph M2 according to the method of step 2); 4) selecting columns and their corresponding attributes in a comparison window according to the content to be compared and visualized, as well as the frequency or time performance to be displayed on the directed edge, so as to generate a process model and statistical information capable of identifying the difference between the two; 5) exporting the process model and comparison information showing the difference in step 4) into PDF format for storage. The present invention uses a flowchart algorithm to visualize and analyze the multi-dimensional attributes between process variants, and can provide guidance for subsequent process modeling and bottleneck analysis and other practical processes.

Description

Technical Field

[0001] The present invention relates to the field of process mining technology, and in particular to a method and system for analyzing multi-dimensional process variant differences based on event logs. Background Art

[0002] Market competition has prompted companies to change their traditional management and operation methods and establish information system management mechanisms within the company to support process mining and help companies provide new monitoring and improvement methods to better adapt to and meet the ever-changing market demands.

[0003] Process mining allows businesses to monitor and optimize their processes. By comparing process behavior under different conditions, such as time and other parameters, it helps identify the causes of differences in model and process performance. Furthermore, comparing the behavior of two processes can help identify why one business process performs better than another based on predefined characteristics or performance metrics, and extract patterns or additional information to improve the performance of the other business process.

[0004] There are two types of process comparison: model-based and log-based. Model-based comparison uses models as input, first generating models from event logs using process discovery techniques, and then examining which activities are present in all models or absent from one model. Model-based comparison is primarily based on the structure of the input models and cannot analyze other process metrics (such as frequency or time performance), while log-based comparison does not have this limitation.

[0005] In order to visualize multiple business processes under specific conditions to find process behaviors and other statistical information with significant differences, potential improvements lie in the innovation of a general technology that can compare process behaviors at the level of multi-dimensional attributes such as performance or frequency, providing guidance for subsequent practical processes such as business process modeling and bottleneck analysis. Summary of the Invention

[0006] The first purpose of the present invention is to overcome the shortcomings and deficiencies of the existing technology and propose a multi-dimensional process variant difference analysis method based on event logs, which can use flowchart algorithms to visualize and analyze the multi-dimensional attributes between process variants, laying the foundation for the management and optimization of subsequent business processes.

[0007] A second object of the present invention is to provide a multi-dimensional process variant difference analysis system based on event logs.

[0008] The first object of the present invention is achieved by the following technical solution: a multi-dimensional process variant difference analysis method based on event logs, comprising the following steps:

[0009] 1) Obtain standardized event logs L1 and L2, which record relevant business processes, including traces, activities, events, resources, and timestamp information;

[0010] 2) Taking the event log L1 as input, the flowchart algorithm is used to transform the event log L1 into a process model that can clearly show the execution path of the business process, specifically a direct follow-up graph M1. The direct follow-up graph is the simplest representation of the process model.

[0011] 3) Add event log L2 that needs to be compared with event log L1, and use the flowchart algorithm to convert event log L2 into a process model that can clearly show the execution path of the business process, specifically directly following figure M2;

[0012] 4) In the process comparison window, select columns and their corresponding attributes according to the content to be compared and visualized, and then filter the variants that meet the conditions in the two event logs L1 and L2. Select the type of value to be displayed on the directed edge, which can be frequency or time performance, to generate a visual process model that can identify the activities and activity edges with significant differences between the two, and display the statistical information of the two, namely the number of trajectories, number of instances and average running time information of the variants;

[0013] 5) Export the visual process model showing the differences and the complete comparison information in step 4) into PDF format.

[0014] Furthermore, in step 1), all event logs are obtained through the 4TU platform.

[0015] Furthermore, in step 2), in the direct follow-up graph, each node represents an activity, and directed edges describe the direct follow-up relationships between various activities. The specific steps are as follows:

[0016] 2.1) Taking the event log L1 as input, set the three parameters of the flowchart algorithm, namely τ var , τ act , τ df , where τ var Defined as the minimum trajectory number threshold for each variant, τ act Defined as the minimum number of events per activity to include, τ df Defined as the minimum number of direct follows for each relation included;

[0017] 2.2) Remove all trajectories with a frequency lower than τ from the event log L1 var Instance of , so that it constitutes a new event log L1 ';

[0018] 2.3) Remove all events with a frequency lower than τ from the new event log L1' actevents, so that it constitutes a new event log L1";

[0019] 2.4) Add a node for each remaining activity in the event log L1' and connect the nodes that satisfy τ df The threshold node is if and only if L1”(a,b)≥τ df When , activity a and activity b are connected, and the directed edge connecting activity a and activity b indicates that activity a in the event log L1”(a,b) is directly followed by activity b;

[0020] 2.5) Output the direct follow graph M1, where nodes are represented by activity frequencies L1”(a), and directed edges are represented by activity frequencies L1”(a, b) with direct follow relationships. Nodes and directed edges can also be represented by time information.

[0021] Further, the step 4) includes the following steps:

[0022] 4.1) Based on the content to be compared and visualized, in the process comparison window, specifically select columns and their corresponding attribute values ​​in the two event logs L1 and L2 to filter the generated process model, and select to filter out other values ​​except the selected values. Columns related to the input event logs can be timestamps, resources, instance names, or activity names, and the attribute values ​​correspond to the selected column values;

[0023] 4.2) Select the type of value to be visualized on the directed edge, which can show the frequency or time performance of the activity, and generate a visualization process model that can identify activities and activity edges that are significantly different between the two. In this case, grayscale lines are used to mark uncommon activities in the process model compared with the compared process model, and the thickness of the directed edge is used to represent the frequency of activity relationships in the event log. Calculate and display statistical information of the two variants, namely the number of trajectories, number of instances, and average running time information of the variants.

[0024] The second objective of the present invention is achieved through the following technical solution: a multi-dimensional process variant difference analysis system based on event logs, comprising:

[0025] A log acquisition module is used to obtain standardized event logs, including trace, activity, event, resource, and timestamp attribute information; wherein, the event log is essentially a finite set of activity sequences that records the execution of business operations;

[0026] The process representation module represents the activity follow-up relationship in the log using a directed graph, i.e., a direct follow-up graph. The activity follow-up relationship depends on the order in which the activities occur, and is represented by directed arrows in the directed graph.

[0027] In the process comparison module, in the process comparison window, you can select columns and corresponding attributes based on the content to be compared and visualized to filter the variants that meet the conditions, and select the type of values ​​to be displayed on the directed edges to generate a visual process model that can identify activities and activity edges with significant differences between the two, and display statistical information for both.

[0028] The result export module exports the visual process model showing the differences and the complete comparison information based on the process comparison results.

[0029] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0030] 1. This invention uses a log-based comparison tool to visualize business processes under specific conditions. Compared with traditional methods, it provides a multi-attribute perspective comparison, thereby providing business process managers with a more favorable decision-making basis.

[0031] 2. The present invention uses a flowchart algorithm to perform difference analysis on the multi-dimensional attributes between process variants, and uses color marking or arc thickness to highlight the differences, so that users can quickly view the different behaviors of two event logs.

[0032] 3. The present invention is highly interactive and can select relevant values ​​and attributes in the event log for comparison and visualization based on the user's usage intention, helping users to accurately analyze the process.

[0033] 4. The present invention can help business process managers view useful knowledge without increasing computing costs, and provide guidance for subsequent business process modeling, bottleneck analysis and other practical processes.

[0034] 5. The present invention has a wide range of applications in business process management and optimization tasks, is simple to operate, and has strong scalability. It has broad prospects in comparative process mining. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 Schematic diagram of the logic flow of the method of the present invention.

[0036] Figure 2 Analytical comparison diagram of the time performance differences between the variants of the present invention.

[0037] Figure 3 Analytical comparison diagram of the frequency differences between variants of the present invention.

[0038] Figure 4 This is a system architecture diagram of the present invention. DETAILED DESCRIPTION

[0039] The present invention will be described in further detail below with reference to the embodiments and drawings, but the embodiments of the present invention are not limited thereto.

[0040] Example 1

[0041] like Figure 1 As shown, this embodiment discloses a multi-dimensional process variant difference analysis method based on event logs, including the following steps:

[0042] 1) Obtain standardized event logs L1 and L2, which record relevant business processes, including traces, activities, events, resources, and timestamp information;

[0043] In this step, the standardized event logs L1 and L2 are real event logs from a road traffic fine information system. The selected event logs L1 and L2 have different sizes, with L2 containing more actions than L1. The details of the selected event logs are shown in Table 1.

[0044] Table 1. Basic information statistics of event log dataset

[0045]

[0046] 2) Using the event log L1 as input, a flowchart algorithm is used to transform the log L1 into a process model that can clearly show the execution path of the business process. Specifically, it is a direct follow-up graph M1. The direct follow-up graph is the simplest representation of the process model. In the direct follow-up graph, each node represents an activity, and directed edges describe the direct follow-up relationship between various activities. The following steps are included:

[0047] 2.1) Taking the event log L1 as input, set the three parameters of the flowchart algorithm, namely τ var , τ act , τ df , where τ var Defined as the minimum trajectory number threshold for each variant, τ act Defined as the minimum number of events per activity to include, τ df Defined as the minimum number of direct follows for each relation included;

[0048] In the event log L1, the number of variants is 10, and τ var Set to 2 to ensure that too many instances are not deleted, and set τ act Set it to 2 to ensure that the number of events is not too small, and set τ df Set it to 3 to ensure that the model is not too complex;

[0049] 2.2) Traverse and mark all the trajectories in the event log L1. When the trajectory frequency is lower than the set value τ var When , remove the trajectory instance and repeat this process so that the trajectory finally retained constitutes a new event log L1';

[0050] 2.3) Traverse the new event log L1' obtained in step 2.2) and record the events in L1'. When the number of events is less than the set value τ act When , remove the low-frequency event and repeat this process to form a new event log L1";

[0051] 2.4) Traverse the remaining activities in the event log L1", add a node for each activity, and connect the nodes that satisfy τ df The threshold node is if and only if L1”(a,b)≥τ df When , activity a and activity b are connected, and the directed edge connecting activity a and activity b indicates that activity a in the event log L1”(a,b) is directly followed by activity b;

[0052] 2.5) Output a direct follow graph M1 based on the result of step 2.4); wherein nodes are represented by activity frequencies L1" (a), and directed edges are represented by activity frequencies L1" (a, b) with direct follow relationships. Nodes and directed edges can also be represented by time information.

[0053] 3) Add event log L2 that needs to be compared with event log L1, and convert it into a process model that can clearly show the business process execution path according to the same method (flowchart algorithm) in step 2), specifically directly following Figure M2.

[0054] 4) In the process comparison window, select columns and their corresponding attributes based on the content to be compared and visualized, and then filter the variants that meet the conditions in the two event logs. Select the type of value to be displayed on the directed edge, which can be frequency or time performance. This generates a visual process model that can identify activities and activity edges with significant differences between the two, and displays statistical information for both, namely the number of trajectories, number of instances, and average running time information of the variants. The specific steps are as follows:

[0055] 4.1) Based on the content to be compared and visualized, in the Process Comparison window, specifically select columns and their corresponding attribute values ​​from the two event logs to filter the generated process model. You can also choose to filter out other values ​​besides the selected values. Columns related to the input event logs can include timestamps, resources, instance names, activity names, and other information, and the attribute values ​​correspond to the selected column values.

[0056] 4.2) Select the type of value to be visualized on the directed edge. This can show the frequency or time performance of the activity. Generate a visual process model that can identify activities and activity edges that are significantly different between the two. In this case, red lines are used to mark uncommon activities in the process model compared to the compared process model. The thickness of the directed edge represents the frequency of activity relationships in the event log. Calculate and display statistical information for the two variants, namely the number of trajectories, number of instances, and average running time information of the variants.

[0057] Attachment Figure 2 The time performance difference analysis results between the two event logs are shown in the following figure. Figure 3 This is the result of the frequency difference analysis between the two event logs. The grayscale lines mark activities that exist in variant 2 but not in variant 1. The numbers on the directed edges indicate the frequency or duration of the activities. Thicker lines indicate more frequent or longer direct follow-up relationships between activities. For example, in variant 1, "fine" >> "payment" is created, where >> indicates a direct follow-up relationship between activities.

[0058] 5) Based on the visualization results of step 4), the visual process model showing the differences and the complete comparison information can be exported to PDF format for saving.

[0059] Example 2

[0060] like Figure 4 As shown, this embodiment discloses a multi-dimensional process variant difference analysis system based on event logs, including the following functional modules:

[0061] A log acquisition module is used to obtain standardized event logs, including trace, activity, event, resource, and timestamp attribute information; wherein, the event log is essentially a finite set of activity sequences that records the execution of business operations;

[0062] The process representation module represents the activity relationships in the log using a direct follow-up graph. The activity relationships are direct follow-up relationships, which depend on the order in which the activities occur. These relationships are represented by directed arrows in the directed graph. The direct follow-up graph removes nodes and edges by setting frequency thresholds to generate a process model, thereby simplifying the process.

[0063] Process comparison module: In the process comparison window, you can select columns and corresponding attributes based on the content to be visualized and compared to filter variants that meet the conditions, and select the type of value to be displayed on the directed edges (such as frequency or time performance) to generate a visual process model that can identify activities and activity edges with significant differences between the two. In this module, grayscale lines are used to mark uncommon activities in the process model compared with the compared process model, and the thickness of the directed edges is used to indicate the frequency of activity relationships in the event log. Statistical information of the two models is also displayed, such as the number of trajectories, number of instances, and average running time of the variants.

[0064] The result export module exports and saves the visual process model showing the differences and the complete comparison information based on the process comparison results.

[0065] To sum up, after adopting the above scheme, the present invention proposes a multi-dimensional process variant difference analysis method and system based on event logs. Taking event logs as input, the system intuitively displays the behaviors with significant differences between different variants by selecting relevant information and corresponding attributes in the comparison window. This can help business process managers view useful knowledge, thereby providing guidance for subsequent business process modeling and bottleneck analysis and other practical processes. It has practical promotion value and is worthy of promotion.

[0066] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be considered as equivalent replacement methods and are included in the scope of protection of the present invention.

Claims

1. A multi-dimensional process variant difference analysis method based on event logs, characterized in that: The following steps are involved: 1) Obtain standardized event logs L1 and L2, which record relevant business processes, including traces, activities, events, resources, and timestamp information; 2) Taking the event log L1 as input, the flowchart algorithm is used to transform the event log L1 into a process model that can clearly show the execution path of the business process, specifically a direct follow-up graph M1, where the direct follow-up graph is the simplest representation of the process model; In the direct follow-up graph, each node represents an activity, and the directed edges describe the direct follow-up relationship between various activities. The specific steps are as follows: 2.1) Taking event log L1 as input, set the three parameters of the flowchart algorithm, namely 𝜏 var , 𝜏 act , 𝜏 df ,in, 𝜏 var Defined as the minimum trajectory number threshold for each variant, 𝜏 act Defined as the minimum number of events per activity to include, 𝜏 df Defined as the minimum number of direct followers for each relation included; 2.2) Remove from event log L1 all tracks with a frequency lower than 𝜏 var Instance of , so that it constitutes a new event log L1'; 2.3) Remove from the new event log L1' all events with a frequency lower than 𝜏 act events, so that they constitute a new event log L1''; 2.4) Add a node for each remaining activity in the event log L1'' and connect the nodes that satisfy 𝜏 df The threshold node is a node if and only if L1''(𝑎, 𝑏) ≥ 𝜏 df When , activity 𝑎 and activity 𝑏 are connected, and the directed edge connecting activity 𝑎 and activity 𝑏 indicates that activity 𝑎 in the event log L1′′(𝑎, 𝑏) is directly followed by activity 𝑏; 2.5) Output the direct follow graph M1, where nodes are represented by activity frequencies L1''(𝑎), and directed edges are represented by activity frequencies L1''(𝑎, 𝑏) with direct follow relationships; 3) Add event log L2 that needs to be compared with event log L1, and use the flowchart algorithm to transform event log L2 into a process model that can clearly show the execution path of the business process, specifically directly following figure M2; 4) In the process comparison window, select columns and their corresponding attributes according to the content to be compared and visualized, and then filter the variants that meet the conditions in the two event logs L1 and L2, and select the type of value to be displayed on the directed edge, which is frequency or time performance, to generate a visual process model that can identify the activities and activity edges with significant differences between the two, and display the statistical information of the two, that is, the number of trajectories, number of instances and average running time information of the variants, including the following steps: 4.1) According to the content to be compared and visualized, in the process comparison window, select the columns and their corresponding attribute values ​​in the two event logs L1 and L2 to filter the generated process model, and select to filter out other values ​​except the selected values, where the columns related to the input event logs are timestamps, resources, instance names or activity names, and the attribute values ​​correspond to the selected column values; 4.2) Select the type of value to be visualized on the directed edge, which is to show the frequency or time performance of the activity, and generate a visualization process model that can identify the activities and activity edges that are significantly different between the two. In this process, grayscale lines are used to mark the uncommon activities in the process model compared with the compared process model, and the thickness of the directed edge is used to indicate the frequency of the activity relationship in the event log. The statistical information of the two variants, namely the number of trajectories, number of instances, and average running time information of the variants, is calculated and displayed; 5) Export the visual process model showing the differences and the complete comparison information in step 4) to PDF format.

2. The multi-dimensional process variant difference analysis method based on event log according to claim 1 is characterized in that: In step 1), all event logs are obtained through the 4TU platform.

3. A multi-dimensional process variant difference analysis system based on event logs, characterized in that: The method for implementing the multi-dimensional process variant difference analysis based on event logs according to claim 1 or 2 comprises: A log acquisition module is used to acquire standardized event logs, including trace, activity, event, resource and timestamp attribute information; wherein the event log is essentially a finite set of activity sequences that records the execution of business operations; The process representation module represents the activity follow-up relationship in the log in the form of a directed graph, i.e., a direct follow-up graph, wherein the activity follow-up relationship depends on the order in which the activities occur, and is represented by directed arrows in the directed graph; Process comparison module: In the process comparison window, select columns and corresponding attributes to filter the variants that meet the conditions according to the content to be compared and visualized, and select the type of values ​​to be displayed on the directed edges to generate a visual process model that can identify the activities and activity edges with significant differences between the two, and display the statistical information of the two at the same time; The result export module exports the visual process model showing the differences and the complete comparison information according to the process comparison results.

Images