Teinemaa et al. 2016 — PPM with Structured and Unstructured Data

Integrates text mining with outcome-prediction PPM. Real logs frequently contain free-text fields (emails, customer comments, case notes) that prior PPM methods simply discarded.

Contribution

PPM framework that extracts features from textual payloads via NLP and combines them with sequence-classification features from structured data.
Evaluated on two real-life logs: debt-recovery and lead-to-contract.
Improvements in both accuracy and earliness of correct outcome prediction.

Significance

Extends the data-aware PPM line into the unstructured-data dimension. Foreshadows later LLM-for-BPM work referenced in sources/2024-kampik-large-process-models.

Cited by

sources/2023-riess-temporal-loss-remaining-cycle-time — Riess cites Teinemaa et al. for the related notion of temporal stability in binary-classification PPM models (distinct from his temporal consistency for regression models).
sources/2024-riess-synbps-simulation-framework — cited for PPM benchmark methodology.

Connections

Concepts: concepts/outcome-prediction · concepts/trace-encoding Authors: entities/irene-teinemaa · entities/marlon-dumas · entities/fabrizio-maggi · entities/chiara-di-francescomarino