object-centric process mining

What is object-centric process mining?

Object-centric process mining (OCPM) is a data science technique to help businesses discover, assess, validate and improve interrelated workflows and processes.

Context is important in understanding OCPM. Generally stated, process mining is the convergence of process science and data science.

Process science focuses on business processes, which are complex and often interdependent relationships of objects and events. For example, the act of hiring and onboarding a new employee involves a clear and well-defined business process. Process science relies on the extensive use of process models and simulations to build and refine business procedures, and it employs familiar technologies such as business process management and workflow automation to understand and manage workflows. However, the data developed and collected in the process is often underused or ignored.

By comparison, data science places its focus on the acquisition and use of business data. For example, business data can include quarterly sales figures or product return/failure rates. It is seen in technology such as machine learning and artificial intelligence. However, the data provided often overlooks or is unrelated to the process that generates the data. Data science is typically unconcerned with the process.

Process mining is an effort intended to blend processes and data into a single cohesive view. The concept of process mining first appeared in the late 1990s as a means of correlating process science and data science. One early expression of process mining was event-based process discovery. These early forays into process mining helped businesses understand the reality of how businesses were operating. As the discipline evolved, businesses used process mining in other efforts such as compliance, forecasting and decision-making.

Process mining is centered around the idea of business objects. Object types can be almost any data, such as sales orders, inventory, accounts payable or order items. These data objects are combined with or compared against business events. Business events are often recorded in an event log. When object data is analyzed in the context of event data, the business can begin to recognize cause-and-event relationships in how the business really operates – and recognize problems or areas for improvement.

There are limitations to traditional process mining, and it is not capable of relating multiple objects. Object-centric process mining seeks to overcome these limitations, letting process analytics capture complex (one-to-many and many-to-many) relationships between business objects. OCPM insights can expand process mining capabilities and cause-and-effect relationships impacting business operations. The relationships between objects are captured in object-centric event logs. These logs can provide detailed analytics and build sophisticated models of the business and its behavior.

Traditional process mining vs. object-centric process mining

Traditional process mining and object-centric process mining share the same fundamental goal: to understand and improve business processes. As a technology, however, OCPM offers greater business insights and benefits because of how it treats business data such as objects and events.

The primary limitation of traditional process mining is a lack of context or depth. A process may be comprised of many individual events, but a given event may be involved in different processes. While it is possible to parse the event involved in a specific process, it requires repetitious and time-consuming data manipulation, which makes process mining more difficult to use.

In practical terms, when an event is related to multiple processes, it's difficult to determine just which process is generating or experiencing problems with an event. This is called convergence. For example, if the same event is triggered for order entry and order shipping, it can be unclear where events correlate to different processes.

Conversely, a phenomenon called divergence can occur when a process might involve differing events. The process might contain variations that sometimes execute or exchange an event with an alternative event.

Where traditional process mining tends to look at and analyze a single object, such as a sales order, OCPM can observe and incorporate the behaviors of related objects. For example, a sales order may also involve objects such as production orders, financial invoices, purchase orders and other types of objects depending on the business. These orders may only be present under certain conditions.

For example, purchase and production orders might only be present if the items represented by a sales order are already in inventory. If not, the items must be produced or procured. At the same time, items in inventory can be shipped, while items that must be produced should be shipped later. All of this affects revenue and financial reporting tasks.

This is a simple example, but it's clear how these variations can add dramatic complexity to the business workflow which traditional process mining tools and techniques cannot see. However, OCPM tools can collect and process these kinds of extended data for a better visualization of the overall process. OCPM works to achieve these capabilities by gathering more data about objects and events and building an object-centric event log to capture interrelated behaviors.

Benefits of object-centric process mining

Object-centric process mining is designed to overcome the limitations of traditional process mining by broadening the ways that event data is collected and processed. This lets business users see events and objects in all their varied interactions along different processes. OCPM can bring several benefits over traditional process mining:

• Better clarity. OCPM can track each object through the event-driven process from start to finish, adding the context of multiple object relationships and reducing the adverse impacts of convergence and divergence. This helps the business better understand delays and make process optimizations.
• Better representations. When process mining is represented graphically, the results can be complex and cumbersome spaghetti charts which are almost impossible to follow. OCPM visualizations are far cleaner and more direct, making it easier for business leaders to understand.
• Better metrics. Metrics are not new in business processes, but the convergence and divergence found in traditional process mining often make metrics difficult to apply. The ability to better associate objects and events within processes makes metrics more meaningful and useful.
• Better context. With greater clarity in objects and relationships, businesses can enhance their understanding of why a process works and why deviations to the process occur. Businesses can then look for ways to improve or standardize the process.
• Better automation. By gathering better insight into the process and its constituent objects and events, the business can streamline processes with fewer exceptions or deviations, allowing for better automation and orchestration initiatives to benefit business automation.

Object-centric process mining tools

There are currently numerous software tools and platforms designed for advanced process mining and business analytics:

• ABBYY Timeline.
• Appian.
• Apromore.
• BusinessOptix.
• Celonis Execution Management System.
• Fluxicon Disco.
• IBM Operational Decision Manager.
• Mehrwerk.
• Microsoft Minit.
• Pegasystems Everflow.
• QPR ProcessAnalyzer.
• SAP Signavio.
• Software AG Aris.
• StereoLogic.
• UIPath.

As with most enterprise-class software products, each offering can focus on different business types, provide diverse analytical capabilities or create certain interoperability limitations. It's important for a business to test and evaluate potential OCPM tools before committing to any one product or vendor.