Skip to main content

Perspectives and event logs

Perspectives isolate specific object types, event types, and relationships from an Object-Centric Process Mining (OCPM) data model. This creates a targeted, analytical view of your process data without modifying the underlying global data structure.

Data analysts use perspectives to subset complex data networks into operational views tailored to specific business metrics. For example, a procurement analyst can create a perspective containing only Purchase Order, Vendor, and Invoice objects to audit late payments within an Order-to-Cash lifecycle, hiding unrelated logistics and inventory layers.

The following interface shows an active perspective tracking duplicate invoice records within an Accounts Payable flow:

An example of a duplicate invoice checker perspective

Key components of perspectives

When creating or using a perspective, you have the following key components:

Component

Purpose

Configuration rule

Object types

Defines the foundational process entities (e.g., Order, Invoice, Customer) included in the view.

Designate exactly one object type as the primary analytical starting point.

Object-to-object relationships

Defines structural connections that allow events to be traced across separate objects.

Include only one explicit relationship path per unique object pair.

Embedded objects

Serves as context-only copies of object types to provide structural detail without expanding the relational graph.

Use embedded objects to break relationship cycles (closed loops) within the model.

Event logs

Organizes the multi-object dataset into a chronological sequence centered around a target lead object.

Mark exactly one event log configuration as the system default.

Graph

Provides the interactive visual mapping canvas to build, evaluate, and validate the perspective layout.

Displays active connections, embedded items, and highlights disconnected object clusters or cycles.

Event logs created from perspectives

Some Celonis apps, such as the Supply Chain Network Visibility App and object-centric Starter Kits, work directly with perspectives. Other apps, including Process Explorer, Process Adherence Manager, and case-centric Starter Kits, use event logs that are created from perspectives.

Event logs reorganize the data from a perspective around a single object. They consolidate all events related to that object into a linear chronological sequence to track the complete operational lifecycle.

When working with event logs:

  • System-generated event logs use the el_ prefix and operate immediately without extra configuration. Advanced analytical variations require custom event logs, which use the el__ prefix.

  • Assign any valid object type as the lead object. The platform supports multiple distinct event logs for a single object type.

  • Filter and subset events by applying specific object attributes. The system outputs additional analytical parameters to the _ActivityDetails columns.

  • Create a custom event log and mark it as the default for applications or features that mandate a single default event log. System-generated event logs cannot serve as the default.

  • Event logs populate activity tables in Studio automatically. Analyze these datasets via views, PQL, or generate them dynamically using the CREATE_EVENTLOG statement.

  • To target performance issues in an active application, override default tracking by initializing a custom log. For instance, call CREATE_EVENTLOG with Lead_Object = el__procurement_compliance to isolate and sequence purchase orders with cycle delays exceeding 14 days.

  • Isolate structural lifecycle deviations by calling downstream string attributes in PQL. For example, query the _ActivityDetails column filtering by Vendor_Region = 'APAC' to verify that localized operational variants match regional governance profiles.

Related topics