The 2024 publications of the Process and Data Science (PADS) group show how rapidly process mining is evolving. Several developments stand out: Object-Centric Process Mining is becoming increasingly central, process mining and generative AI are converging, behavioral analysis is becoming richer, and process intelligence is emerging as an important foundation for enterprise AI.
One of the clearest trends is the transition from traditional case-centric process mining toward Object-Centric Process Mining (OCPM). Classical event logs assume that each event belongs to one case. However, real operational processes typically involve many interacting objects, such as orders, items, invoices, deliveries, machines, patients, and treatments.
The OCEL 2.0 specification provides an important foundation for representing such object-centric event data. At the same time, object-centric techniques are being extended to an increasingly broad range of analysis tasks, including predictive monitoring, conformance checking, performance analysis, anomaly detection, and process comparison. An example is Improving Predictive Process Monitoring Using Object-Centric Process Mining.
This development reflects a more fundamental shift: instead of forcing operational reality into a single case notion, process mining increasingly aims to capture the actual interconnected structure of processes.
A second major trend is the rapidly growing interaction between process mining and Large Language Models (LLMs). Several papers investigate how generative AI can support process modeling, discovery, interpretation, and data preparation.
ProMoAI, for example, investigates the automatic generation of process models from natural-language descriptions. The paper Bridging Domain Knowledge and Process Discovery Using Large Language Models explores how domain knowledge expressed through LLMs can complement purely data-driven process discovery.
At the same time, it is important not to simply assume that LLMs understand processes well. The PM-LLM Benchmark therefore focuses on systematically assessing the capabilities and limitations of LLMs on process-mining-related tasks.
The relationship between AI and process mining also works in the opposite direction. AI systems need reliable context about how organizations actually operate. Event data and process models can provide such context and help AI systems reason about constraints, dependencies, and likely consequences of actions.
The paper Incorporating Behavioral Recommendations Mined from Event Logs into AI Planning illustrates this direction by connecting behavioral knowledge extracted from event data with AI planning.
This is an increasingly important theme: AI can support process mining, but process mining can also make AI more grounded, explainable, and operationally relevant.
The 2024 publications also show continued work on richer representations of process behavior. Topics include stochastic behavior, partial orders, long-term dependencies, declarative constraints, conformance diagnostics, behavioral patterns, and high-level events.
This work recognizes that real processes cannot always be understood through a single deterministic flow diagram. Events may occur concurrently, probabilities matter, dependencies may stretch over long periods, and the same operational process can exhibit many different forms of behavior.
Another recurring theme is the development of scalable software and reusable research infrastructure. Rust4PM, for example, explores high-performance process-mining implementations while making functionality accessible from established programming environments.
The range of applications also continues to expand. The 2024 publications address domains including healthcare, manufacturing, cyber-physical systems, purchase-to-pay processes, mining, education, scientific workflows, and high-performance computing. This breadth illustrates how process mining is developing from a specialized analysis technique into a general approach for understanding operational systems.
Taken together, the 2024 publications point toward a broader transformation of the field: from cases to objects, from isolated process models to rich behavioral representations, and from retrospective analysis toward prediction, decision support, and intelligent action.
These developments increasingly come together under the notion of Process Intelligence. Object-centric event data provides the context, process mining transforms this data into operational knowledge, and AI can use this knowledge to reason and act.
The emerging message is therefore not AI instead of process mining, but AI grounded in process intelligence.
See the complete overview of the 2024 publications.