---
title: "Process Mining 2.0: From Insights to Action"
type: source
tags: [process-mining, keynote, prescriptive-monitoring, apromore, vision]
authors: [Dumas Marlon]
year: 2021
venue: "AI4EPT workshop @ PAKDD 2021 (keynote slides)"
kind: presentation
raw_path: "raw/Predictive process monitoring/Dumas - from insights to action.pdf"
created: 2026-04-13
updated: 2026-04-13
key_claims:
  - Process Mining 1.0 stopped at insights; Process Mining 2.0 must drive action.
  - Prescriptive process monitoring recommends interventions, not just predictions.
  - The loop from mining to action passes through prediction, causal inference, and intervention evaluation.
---

# Dumas 2021 — Process Mining 2.0: From Insights to Action (Keynote)

Keynote slides by [[entities/marlon-dumas]] delivered at the AI4EPT workshop at PAKDD 2021. Not a peer-reviewed paper — a vision presentation.

## Thesis
Process Mining 1.0 stops at **insights** (a discovered model, annotated with performance). Process Mining 2.0 must close the loop to **action** — intervention recommendations, causal reasoning about what-if scenarios, and measurable business impact.

## Relation to broader literature
The argument is the conceptual predecessor of the *Augmented Process Execution* paper ([[sources/2023-chapela-campa-augmented-process-execution]]) and foreshadows the [[concepts/conversational-actionability|Recommend]] enactment role in APM ([[sources/2026-calvanese-agentic-bpm-manifesto]]). Sits at the **prescriptive** end of the descriptive→predictive→prescriptive analytics pyramid.

Siblings in the prescriptive space:
- [[sources/2014-groger-prescriptive-analytics-bpo]] — Gröger, Schwarz & Mitschang's early rBPO paper (BIS 2014) is the **predecessor** to Dumas's vision: a concrete architecture (process warehouse + decision-tree prediction + decision-tree recommendation) for prescriptive analytics in BPM, preceding this keynote by 7 years.
- [[sources/2022-kubrak-prescriptive-ppm-slr]] — Kubrak, Milani, Nolte & Dumas's PrPM SLR (2022) is the **formalisation** of the field this keynote called for: it catalogues 37 methods along six dimensions and names the research gaps (in-vivo validation, causal policies, terminology).

## Philosophical bite
Closing the loop from prediction to action smuggles in a strong causal commitment: the recommended intervention must *cause* the improved outcome, not merely correlate with it. This is an [[concepts/interventionist-theory-of-causation|interventionist]] claim and carries all the caveats the PHI403 course raises about [[concepts/rct-limitations|external validity]], [[concepts/mechanisms-in-science|mechanism]], and individual-vs-population [[concepts/probabilistic-causation|propensities]] — see [[sources/2023-anjum-rocca-phi403-causation-in-science]].

## Connections
**Concepts:** [[methods/process-mining-basics]] · [[concepts/predictive-process-monitoring]] · [[concepts/prescriptive-process-monitoring]] · [[concepts/conversational-actionability]] · [[concepts/intervention-policy]] · [[concepts/interventionist-theory-of-causation]] · [[concepts/causation]]
**Authors:** [[entities/marlon-dumas]]
**Related sources:** [[sources/2014-groger-prescriptive-analytics-bpo]] · [[sources/2022-kubrak-prescriptive-ppm-slr]]