Object-Centric Process Mining (OCPM)

Classical process mining has produced transformative insights for thousands of organizations over the past two decades. But it rests on a simplifying assumption that does not hold for many real business processes: that every event belongs to exactly one case, identified by a single case ID.

In practice, business processes — particularly in ERP environments — involve many-to-many relationships. A single invoice may cover multiple purchase orders. A single purchase order may contain multiple line items, each received separately. A single customer may have dozens of simultaneous open orders processed by overlapping sets of employees. Object-Centric Process Mining (OCPM) was developed to handle this reality without the distortions that classical process mining introduces when forced into a single case-ID model.

The Problem with Single Case IDs

When a process involves many-to-many relationships and classical mining is applied, one of two distortions occurs:

• Convergence (many-to-one): Multiple case identifiers are merged into one. An invoice that covers 12 purchase orders is treated as a single case — hiding the individual purchase order flows and misrepresenting the process structure. The resulting process map appears simpler than reality.
• Divergence (one-to-many): A single event is replicated across multiple cases. A goods receipt event involving one shipment covering five purchase orders is copied five times — once for each PO. The duplicated events create artificial loops and inflated frequencies in the process map, significantly distorting performance metrics like cycle time and frequency.

These distortions are not marginal edge cases — they affect the most important and highest-volume processes in most ERP systems: Purchase-to-Pay, Order-to-Cash, inventory management and financial close.

How OCPM Works

OCPM replaces the flat event log structure of classical mining with an Object-Centric Event Log (OCEL). An OCEL associates each event with multiple objects simultaneously:

• An event "Invoice Posted" may be associated with objects: Invoice #INV-4421, Purchase Orders #PO-1001 and #PO-1002, Vendor #V-0045 and Ledger Account #4000.
• Each object type (Invoice, Purchase Order, Vendor, Account) has its own attributes and lifecycle.
• The OCEL captures the relationships between objects at the event level, preserving the many-to-many structure of real processes.

OCPM algorithms then discover process models that reflect this multi-object reality — showing how objects interact, where they converge and diverge, and what the process looks like from each object type's perspective simultaneously.

OCEL 2.0: The Emerging Standard

The Object-Centric Event Log standard has been formalized through the IEEE Task Force on Process Mining. OCEL 2.0, published in 2023, defines a comprehensive data model including events, objects, object types, attributes, relationships and qualifiers. The standard is supported by open-source libraries (ocel2-pm4py) and is being incorporated into commercial process mining tools.

For organizations considering process mining tooling investments, OCEL 2.0 support is an increasingly important vendor selection criterion — particularly for ERP-based use cases where many-to-many process structures are the norm rather than the exception.

OCPM and Business Central

Business Central's data model is inherently object-centric. A sales order header relates to multiple sales lines, each relating to an item, a location, a shipment and ultimately an invoice. A purchase order relates to purchase lines, receipts, vendor invoices and payment entries. This relational structure maps directly to the OCEL data model — making BC data particularly well-suited to OCPM analysis.

Classical process mining of BC data requires choosing one object type as the case identifier and accepting the resulting distortions. OCPM allows all relevant object types to be modeled simultaneously, producing a more accurate and complete view of how processes actually behave across the full relational structure of the ERP. As OCPM tooling matures and OCEL 2.0 support becomes standard in process mining extensions, BC users will be able to access significantly richer process analysis than classical mining permits.

Current Maturity and Practical Availability

OCPM is an active area of research and tooling development. As of 2026:

• The theoretical foundations and key algorithms are well-established in academic literature
• Open-source libraries (PM4Py with OCEL support) allow experimentation
• Several commercial vendors have announced or begun releasing OCEL-compatible features
• Enterprise-grade, production-ready OCPM tooling is still emerging for most vendors

Organizations evaluating process mining investments should track OCPM capability in their vendor shortlists — not necessarily as an immediate requirement, but as an indicator of vendor research alignment and future roadmap direction.

Related Concepts

• Process Mining
• Process Mining Algorithms
• AI Process Mining
• Event Log
• Conformance Checking
• Purchase-to-Pay
• Order-to-Cash