Correction to Large Process Models — Kampik et al. 2024

Two-page erratum (Springer, online 19 December 2024) to the original LPM paper. The published Figure 1 had characters missing from several block labels; the correction restores them. The text of the article is unaffected.

What was wrong

In the original Fig. 1:

"Process Knowled e & Expertise" — missing 'g' in "Knowledge".
"Insi hts, KPIs, Live Data, Experience Data" — missing 'g' in "Insights".
"Process Prediction Engines (Classical & dation Models)" — missing "Foun" prefix in "Foundation Models".
The Process Fine-tuned LLM sub-block label inside the LLM region was absent.
The three boxes inside BPM Tools & Integration (Process Modelling; Process Analysis/Mining; Process Execution) were not labelled.

What the corrected figure shows

The architecture is unchanged; the corrected Fig. 1 makes explicit:

Top row (Process Data and Knowledge Sources): Process Knowledge & Expertise · Process/Customer Context · Insights, KPIs, Live Data, Experience Data — feeding into the Process Atom Layer.
Foundation-model region (orange): Curated Process Knowledge + General Purpose LLM (Foundation Model), wrapped together as a Process Fine-tuned LLM.
Run-time contextualization (blue): Process Query Engines + Process Prediction Engines (Classical & Foundation Models).
BPM Tools & Integration (red): Process Modelling · Process Analysis/Mining · Process Execution.
Feedback loop (right): human/machine process manager — Specify BPM Goal → Consume Recommendation/Action → Provide Feedback.

Connections

Corrects: sources/2024-kampik-large-process-models