Probabilistic Inductive Miner for Process Tree Discovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing process model discovery techniques generate complex process models that are not usable in many applications due to their complexity, despite being capable of expressing complex behaviors like parallelism.
Innovation Solution
A probabilistic inductive miner system that recursively generates process trees by splitting event logs into sub-event logs based on frequency of directly and indirectly follows relations, adding nodes representing relationship operators, and determining cut locations to simplify the process tree structure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If existing process model discovery techniques are used to generate formal process models, then complex behaviors such as parallelism can be expressed, but the process models become too complex to be utilized in many applications
Solution Approach 1:
The patent segments the event log into multiple sub-event logs based on frequency thresholds of directly follows and indirectly follows relations. This segmentation allows the discovery algorithm to process smaller, more manageable subsets of the data independently, resulting in simpler process models that can be later composed together to represent the complete process behavior including parallelism and other complex patterns.
2Ease of operation
If process models are simplified for usability, then they become easier to utilize, but they may lose the ability to express complex behaviors accurately
Solution Approach 1:
The patent merges multiple simple process models derived from segmented sub-event logs into a composite process model that accurately represents the complete process behavior. By combining the results from multiple simplified analyses with frequency-based relationship tracking, the system achieves both simplicity in individual components and accuracy in the overall model, preserving complex behaviors like parallelism through the composition of simpler elements.
Data Source
AI summary
Systems and methods for generating a process tree of a process are provided. An event log of the process is received. It is determined whether a base case applies to the event log and, in response to determining that the base case applies to the event log, one or more nodes are added to the process tree. In response to determining that the base case does not apply to the event log, the event log is split into sub-event logs and one or more nodes are added to the process tree. The steps of determining whether a base case applies and splitting the event log are repeatedly performed for each respective sub-event log using the respective sub-event log as the event log until it is determined that the base case applies to the event log. The process tree is output. The process may be a robotic process automation process.