Unlocking Non-Block-Structured Decisions: Inductive Mining with Choice Graphs

Abstract: Process discovery aims to automatically derive process models from event logs, enabling organizations to analyze and improve their operational processes. Inductive mining algorithms, while prioritizing soundness and efficiency through hierarchical modeling languages, often impose a strict block-structured representation. This limits their ability to accurately capture the complexities of real-world processes. While recent advancements like the Partially Ordered Workflow Language (POWL) have addressed the block-structure limitation for concurrency, a significant gap remains in effectively modeling non-block-structured decision points. In this paper, we bridge this gap by proposing an extension of POWL to handle non-block-structured decisions through the introduction of choice graphs. Choice graphs offer a structured yet flexible approach to model complex decision logic within the hierarchical framework of POWL. We present an inductive mining discovery algorithm that uses our extension and preserves the quality guarantees of the inductive mining framework. Our experimental evaluation demonstrates that the discovered models, enriched with choice graphs, more precisely represent the complex decision-making behavior found in real-world processes, without compromising the high scalability inherent in inductive mining techniques.

Unlocking Non-Block-Structured Decisions: Inductive Mining with Choice Graphs

The paper "Unlocking Non-Block-Structured Decisions: Inductive Mining with Choice Graphs" introduces an extension of the Partially Ordered Workflow Language (POWL), which aims to address an enduring limitation in process discovery: the inability to effectively model non-block-structured decision points. This limitation arises due to the strict hierarchical nature of existing inductive mining algorithms, which traditionally impose a block-structured representation of process models. While recent advancements like the Partially Ordered Workflow Language have alleviated constraints around concurrency, complexities surrounding decision points remained a challenge in capturing real-world processes accurately.

Core Contributions

This research bridges the existing gap by proposing "POWL 2.0," which introduces choice graphs—a novel construct designed to encapsulate decision-making logic without adhering to strict block-structures. Choice graphs, inspired by the flexibility of Directly-Follows Graphs (DFGs), offer a structured yet flexible methodology to represent non-block-structured decisions. The significance of choice graphs lies in their ability to circumvent overgeneralization and preserve the high scalability of inductive mining.

The paper also delineates an adaptation of the Inductive Miner, enabling the discovery of POWL 2.0 models. This adaptation involves detecting valid choice graph cuts over event logs, thus embedding decision logic into the mining process effectively. The proposed algorithm is backed by formal proofs that demonstrate the preservation of fitness guarantees native to the Inductive Miner, ensuring that every trace in the event log is representable within the model's language.

Experimental Evaluation

An empirical assessment against real-life event logs indicates that the introduction of choice graphs enhances model precision and f-score relative to the standard POWL Inductive Miner. The results suggest that choice graphs not only simplify the representation of complex branching logic but also improve the overall model quality across various dimensions. Importantly, these improvements are achieved without compromising computational efficiency, as the extended algorithm often maintains or even reduces execution times compared to existing approaches, especially when combined with noise filtering.

Implications and Future Directions

The theoretical advancements and empirical validations presented in the paper have substantial implications for both practical applications and further research. Practically, organizations leveraging process discovery can expect more nuanced and accurate representations of their workflows when utilizing the proposed methodologies. On the theoretical front, the integration of choice graphs into process mining frameworks sets a foundation for expanding POWL with additional constructs to model complex dependencies or non-linear executions.

Moreover, future research could explore optimizing choice graph detection algorithms further, potentially incorporating machine learning techniques to refine edge detection and partitioning strategies. Additionally, the controlled integration of cyclic choice graphs could be explored to handle intertwining decision points within process models more effectively.

By advancing the ability to model non-block-structured decisions, this research contributes significantly to the field of process mining and propels the capabilities of automatic process discovery towards more comprehensive, scalable, and efficient solutions.