arXiv:2608. 09398v1 Announce Type: cross Abstract: Process discovery is one of the central challenges in process mining.
Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.
PetriBench is a compact, fully self‑contained, and scalable benchmark that evaluates large language model (LLM) reasoning over dynamic state spaces using Petri nets. It organizes reasoning into four task families with Easy, Medium, and Hard levels, each generated by increasing structural complexity and evaluated against exact ground truth. Experiments across proprietary and open‑weight models show that accuracy consistently drops with difficulty, revealing distinct task‑specific capability profiles, while test‑time compute and procedural generation affect performance differently across tasks.
arXiv:2607. 04856v1 Announce Type: new Abstract: Local Process Models (LPMs) are an underexplored concept in process mining.
The paper introduces a new approach to process model forecasting that uses causal nets instead of traditional directly-follows graphs, enabling explicit representation of concurrency. It forecasts time series of relation and binding counts, reconstructs future process models with AND/XOR semantics, and evaluates them using a protocol that handles partial traces for conformance checking. Experiments on four event logs show that the forecasted models achieve conformance close to re‑mined models and outperform static discovery baselines, though filtering infrequent bindings improves metrics at the cost of losing concurrent behavior.
The paper examines how event data from hospital processes can obscure whether activities occur concurrently or sequentially. It shows that standard event‑log approaches, based on stochastic language, often fail to distinguish concurrency because any log can be explained by a model with no concurrent events. The authors argue that the key to resolving this ambiguity lies in the choice of what is recorded—such as precise start and end times or object‑centric ordering—rather than simply collecting more data.
The paper examines how event logs used in process mining can fail to reveal concurrent versus sequential activities, using a hospital example where blood tests and imaging may occur simultaneously or in alternating order. It demonstrates that standard stochastic language approaches only expose the assumptions of their discovery algorithms, often misrepresenting concurrency. The authors argue that the key to distinguishing concurrent behavior lies in the choice of recorded data—such as precise start and end times or object‑centric ordering—rather than simply increasing sample size.
arXiv:2604. 22455v2 Announce Type: replace Abstract: A core component of any AI-Augmented Business Process Management System (ABPMS) is the process frame, which gives the system process-awareness and defines its maximal behavioral boundaries.