The WHY in Business Processes: Discovery of Causal Execution Dependencies

Overview

• Discovering the causal relationships between process activities is crucial for predicting the consequences of process interventions and making informed decisions about process improvements.
• Existing process discovery algorithms rely mainly on time precedence, but a causal view can provide a new perspective where relations reflect genuine cause-effect dependencies.
• This paper presents a systematic approach to unveiling the causal business process by leveraging an existing causal discovery algorithm over activity timing.
• The paper also identifies conditions under which process mining discovery algorithms generate models that are incongruent with the causal business process model and shows how the causal model can be used for sound process analysis.

Plain English Explanation

Businesses often have complex processes involving various activities, like approving a loan or manufacturing a product. Understanding how these activities are causally connected - what causes what - is important for making smart decisions about improving the process. Existing methods mainly look at the order of activities to build process models, but this doesn't always capture the true cause-and-effect relationships.

This paper proposes a new approach that uses an algorithm for discovering causal connections, based on the timing of the activities. By applying this causal discovery technique, the researchers can uncover the underlying causal structure of the business process, which can be different from what standard process mining tools would find.

The paper also identifies situations where the standard process models may not align with the true causal relationships. It shows how to use the causal model to spot these discrepancies and get a more accurate understanding of what's really going on in the process. This can lead to better decisions about how to optimize or streamline the process.

The researchers demonstrate their methodology using two open-source process mining tools and test it on some sample datasets. The causal perspective they introduce provides a new way to analyze and improve business processes that goes beyond just looking at the sequence of activities.

Technical Explanation

The paper presents a systematic approach to discovering the causal business process by leveraging a causal discovery algorithm, LiNGAM, over activity timing data. This allows the researchers to uncover genuine cause-effect dependencies among the tasks, which can supplement standard process discovery techniques that rely primarily on time precedence.

The key innovation is the identification of conditions under which process mining discovery algorithms, like the IBM Process Mining tool, generate a model that is incongruent with the causal business process model. The paper shows how the causal model can be methodologically employed to analyze and annotate these inconsistencies in the context of three causal patterns.

The methodology is demonstrated on a synthesized dataset as well as two open benchmark datasets. By comparing the mined process model to the causal model, the researchers are able to surface discrepancies and provide a richer, more accurate view of the underlying process.

This causal perspective represents a departure from traditional process mining, which has tended to focus on temporal precedence. The techniques described in the paper offer a new way to leverage causal discovery for process analysis and improvement, going beyond simply observing the sequence of activities.

Critical Analysis

The paper makes a compelling case for the value of causal modeling in process mining, but there are a few notable limitations and areas for further research:

1. The experiments are conducted on relatively small, synthetic or benchmark datasets. It would be important to validate the approach on larger, real-world business process data to assess its scalability and practical applicability.

2. The paper does not address the robustness of the causal discovery algorithm (LiNGAM) to noisy or incomplete data, which is a common challenge in real-world process mining scenarios. Advances in causal discovery under uncertainty could be explored.

3. The integration of causal discovery with trusted execution environments for decentralized process mining is not discussed, but could be an important consideration for applying these techniques in practice.

4. While the paper identifies discrepancies between the mined process model and the causal model, it does not provide guidance on how to resolve these inconsistencies or determine which model is more accurate. Further research on model reconciliation would be valuable.

Overall, the paper presents a promising new direction for process mining by incorporating causal reasoning, but additional work is needed to fully realize the potential of this approach in real-world business settings.

Conclusion

This paper introduces a novel approach to process mining that goes beyond simply observing the temporal order of activities. By leveraging causal discovery techniques, the researchers can uncover the underlying causal structure of business processes, which may differ from the models generated by standard process mining algorithms.

The ability to identify discrepancies between the mined process model and the causal model provides a richer, more accurate understanding of the process and can lead to better-informed decisions about process improvements. While further research is needed to address some of the limitations, this work represents an important step towards a causal perspective in process mining that can supplement traditional techniques.

Overall, the insights from this paper have the potential to drive more effective and sustainable process optimization, with implications for diverse industries and applications.