Skip to main content

The object-centric data model (OCDM)

An object-centric data model in the Celonis platform represents business processes by modeling all relevant business objects and their relationships in a single, unified structure. Instead of focusing on a single case notion, such as an order or a ticket, the object-centric model captures multiple interconnected objects (for example, orders, deliveries, invoices, and payments) and their events. This approach enables a more realistic and complete view of complex, end-to-end processes, supporting deeper analysis, transparency, and more accurate insights across the process landscape.

An example of the graph view of an object-centric data model:

process_intelligence_graph_example.png

Viewing the object-centric data model

Within Objects and Events, you can view your object-centric data model in the following ways:

The graph view is a visual representation of your object-centric data model. It shows object types as nodes and relationships as lines, making it easier to understand how objects and events are connected across your processes.

To access the graph view from the dashboard, click View Graph:

access_OCDM_graph.png

When viewing the graph, you have the following features:

  • Navigation tools: The Graph view displays object types as nodes and their relationships as lines. You can pan by dragging the gray background and zoom using the controls at the bottom of the model. Drag objects to reposition them, or click Auto align to restore the optimized layout.

    navigation_tools.png
  • Search: Search for any object type or event type and see its details.

    search_ocdm.png
  • Filters: Select Add a filter to create your own combination of filters by namespace (Celonis or custom), tags, or object types. You can only have one active filter of each kind.

    add_a_filter_in_ocdm.png
  • Presets: Presets make it easy to switch between different process views without manually reapplying filters. Click + Save to

    save_presets.png

The object-centric data model can also be viewed as a list of objects and events, accessed by clicking either Objects or Events from the dashboard:

click_objects_and_events.png

When viewing the list, you have the following features:

  • Search: Search for any object type or event type and see its details.

    search_list_view.png
  • Filter: Select an existing category and filter to narrow down the list of objects or events. Click + to add your own custom filter category.

    list_view_filters.png
  • Sort: Sort the list of objects and events based on alphabetical, creation date, or date last modified.

    sort_list_view.png
  • View in Graph: Click View in Graph to view that object or event in the graph view of the OCDM.

    list_view_view_in_graph.png

Comparing the object-centric data model to a case-centric data model

When considering if an object-centric data model makes sense for your use cases, consider the following:

Feature

Object-centric data model

Case-centric data model

Core concept

Multiple business objects interact in one process, giving you a network of connected object lifecycles.

One case ID represents the process instance, showing you a single, linear flow.

Data structure

  • Multiple object tables

  • One or more event tables

  • Events can be linked to multiple objects

  • Object-to-object relationships preserved

  • One event table

  • One case ID

  • Events belong to exactly one case

  • Other entities are flattened into attributes

Handling complexity

  • 1-to-many relationships: Supported.

  • Many-to-many relationships: Supported.

  • Rework / partial flows: Supported.

  • Parallel activities: Supported.

  • 1-to-many relationships: Not supported, the relationships are flattened or duplicated.

  • Many-to-many relationships: Not supported.

  • Rework / partial flows: Results are distorted.

  • Parallel activities: While supported, they're difficult to model effectively.

Data quality and accuracy

Using object-centric data models has some of the following data quality advantages:

  • No duplication

  • Real timestamps

  • True process paths

  • Accurate KPIs (lead time, rework, automation)

When using case-centric data models, you may experience some of the following data quality issues:

  • Event duplication

  • Artificial event ordering

  • Inflated lead times

  • Incorrect conformance results