Causal process discovery and generation of causal process discrepancy views from process event logs
View Patent ↗An example system includes a processor to receive a process discovery model generated based on a process event log using a process discovery process. The processor can generate a causal process model based on the process event log. The processor can generate a causal process discrepancy view based on a comparison of the process discovery model and the causal process model with respect to a causal execution pattern.
1. A system, comprising a processor to:
receive a process discovery model generated based on a process event log using a process discovery process;
extract transposed timestamp tuples adjusted relative to a start time of the process discovery model for pairs of activities from data present in the process event log;
generate a causal process model based on the process event log and the transposed timestamp tuples that corresponds to a confounder structure, wherein the confounder structure includes the data present in the process event log; and
generate a causal process discrepancy view based on a comparison of the process discovery model and the causal process model with respect to a causal execution pattern, wherein the causal process discrepancy view, highlights discrepancies among executions of the pairs of activities from the process event log.
2. The system of claim 1 , wherein the causal execution pattern comprises a collider-based causal execution pattern.
3. The system of claim 1 , wherein the causal execution pattern comprises a confounder-based causal execution pattern.
4. The system of claim 1 , wherein the causal execution pattern comprises a mediator-based causal execution pattern.
5. The system of claim 1 , wherein the processor is to generate a causal annotation for the causal process discrepancy view in response to detecting that the process discovery model matches the causal execution pattern.
6. The system of claim 1 , wherein the processor is to reconstruct the process discovery model to match a true causal execution pattern in response to detecting that the process discovery model does not match any causal execution pattern in the causal process model.
7. The system of claim 1 , wherein the causal process model comprises a process model in which inter-activity relations among any of its activities have a causal execution dependence.
8. The system of claim 1 , wherein the processor is to exclude causal relations that go against process progression from the process discovery model to generate the causal process discrepancy view.
9. A computer-implemented method, comprising:
receiving, via a processor, a process discovery model generated based on a process event log using a process discovery process;
extracting transposed timestamp tuples adjusted relative to a start time of the process discovery model for pairs of activities from data present in the process event log;
generating, via the processor, a causal process model based on the process event log and the transposed timestamp tuples that corresponds to a confounder structure, wherein the confounder structure includes the data present in the process event log; and
generating, via the processor, a causal process discrepancy view based on a comparison of the process discovery model and the causal process model with respect to a causal execution pattern, wherein the causal process discrepancy view, highlights discrepancies among executions of the pairs of activities from the process event log.
10. The computer-implemented method of claim 9 , wherein generating the causal process model based on the process event log comprises:
executing a causal discovery process over the transposed timestamp tuples to determine if either activity in each of the pairs of activities causes another activity; and
removing all pairs in response to detecting that an opposite direction pair exists in the process discovery model.
11. The computer-implemented method of claim 9 , wherein the causal execution pattern comprises a collider-based causal execution pattern, further comprising, for each sequence of three activities from the process discovery model that is detected as being a collider structure with a third activity being a collider, annotating a first relation between a first activity and a second activity as not causal, annotating a second relation between the second activity and the third activity as causal, and annotating a third relation between the first activity and the third activity as causal and as not included as a time precedence relation in the process discovery model.
12. The computer-implemented method of claim 9 , wherein the causal execution pattern comprises a confounder-based causal execution pattern, further comprising, for each sequence of three activities from the process discovery model that is detected as being a confounder structure with a first activity being a confounder, annotating a first relation between a second activity and a third activity as not causal, annotating a second relation between the first activity and the second activity as causal, and annotating a third relation between the first activity and the third activity as causal and as not included as a time precedence relation in the process discovery model.
13. The computer-implemented method of claim 9 , wherein the causal execution pattern comprises a mediator-based causal execution pattern, further comprising, for each sequence of three activities from the process discovery model that is detected as being a mediator structure with a second activity being a mediator, annotating a first relation between a first activity and the second activity as causal, annotating a second relation between the second activity and a third activity as causal, and annotating a third relation between the first activity and the third activity as causal and as not included as a time precedence relation in the process discovery model.
14. The computer-implemented method of claim 9 , wherein generating the causal process discrepancy view comprises blending results from a comparison of the process discovery model and the causal process model with respect to a collider-based causal execution pattern, a confounder-based causal execution pattern, and a mediator-based causal execution pattern into the causal process discrepancy view.
15. A computer program product for generating causal process discrepancy views, the computer program product comprising a computer-readable storage medium having program code embodied therewith, the program code executable by a processor to cause the processor to:
receive a process discovery model generated based on a process event log using a process discovery process;
extract transposed timestamp tuples adjusted relative to a start time of the process discovery model for pairs of activities from data present in the process event log;
generate a causal process model based on the process event log and the transposed timestamp tuples that corresponds to a confounder structure, wherein the confounder structure includes the data present in the process event log; and
generate a causal process discrepancy view based on a comparison of the process discovery model and the causal process model with respect to a causal execution pattern, wherein the causal process discrepancy view, highlights discrepancies among executions of the pairs of activities from the process event log.
16. The computer program product of claim 15 , further comprising program code executable by the processor to:
execute a causal discovery process over the transposed timestamp tuples to determine if either activity in each of the pairs of activities causes another activity; and
remove all pairs in response to detecting that an opposite direction pair exists in the process discovery model.
17. The computer program product of claim 15 , wherein the causal execution pattern comprises a collider-based causal execution pattern, further comprising program code executable by the processor to:
annotate, for each sequence of three activities from the process discovery model that is detected as being a collider structure with a third activity being a collider, a first relation between a first activity and a second activity as not causal;
annotate a second relation between the second activity and the third activity as causal; and
annotate a third relation between the first activity and the third activity as causal and as not included as a time precedence relation in the process discovery model.
18. The computer program product of claim 15 , wherein the causal execution pattern comprises a confounder-based causal execution pattern, further comprising program code executable by the processor to:
annotate, for each sequence of three activities from the process discovery model that is detected as being a confounder structure with a first activity being a confounder, a first relation between a second activity and a third activity as not causal;
annotate a second relation between the first activity and the second activity as causal; and
annotate a third relation between the first activity and the third activity as causal and as not included as a time precedence relation in the process discovery model.
19. The computer program product of claim 15 , wherein the causal execution pattern comprises a mediator-based causal execution pattern, further comprising program code executable by the processor to:
annotate, for each sequence of three activities from the process discovery model that is detected as being a mediator structure with a second activity being a mediator, a first relation between a first activity and the second activity as causal;
annotate a second relation between the second activity and a third activity as causal; and
annotate a third relation between the first activity and the third activity as causal and as not included as a time precedence relation in the process discovery model.
20. The computer program product of claim 15 , wherein program code executable by the processor to generate the causal process discrepancy view further comprises program code executable by the processor to:
blend results from a comparison of the process discovery model and the causal process model with respect to a collider-based causal execution pattern, a confounder-based causal execution pattern, and a mediator-based causal execution pattern into the causal process discrepancy view.