Discovery and Simulation of Business Process with Multiple Data Attributes and Conditions
Name
Serhii Murashko
Abstract
Business process simulation (BPS) is a crucial tool for organizations, allowing them to forecast outcomes and assess the impact of potential changes within their processes. This capability supports effective decision-making by enabling "what-if" scenario analysis. However, traditional BPS models rely on a limited number of attributes (activity names, resources, and timestamps) with probabilistic decision-making, overlooking process-dependent data attributes. For example, in emergency services or patient care processes, using probabilistic decisions might miss important details about patient conditions or available resources, which could trigger critical issues.
This thesis introduces a Data-Aware Simulation (DAS) model, designed to incorporate dynamic attributes and enable data-aware decisions within simulations. The DAS model addresses the limitations of traditional approaches categorising attributes into 3 types, case, global, and event, to cover different scopes (local or global) and behaviour (static or dynamic) of the attributes. These attributes provide branching conditions at the decision points based on the current data state to guide the execution flow of the process.
Another significant aspect of this research is the discovery of the DAS model from the event logs. By incorporating the DAS model with discovery tools, organisations can discover their simulation models, including the perspective of data attributes and branching conditions that often affect the execution flow of business processes. Following discovery, organisations can simulate the model and make necessary adjustments and optimisations to adapt the models to reflect changes in operational procedures or to explore different ’what-if’ scenarios, thereby maintaining their relevance and effectiveness in dynamic business environments.
The evaluation demonstrates that data-aware models, discovered from event logs, can accurately classify data attributes, their update mechanisms, and implications into branching conditions. These models also replicate the control flow of the original log while enhancing cycle and event times, in contrast to traditional non-data-aware models that depend on branching probabilities.
This thesis introduces a Data-Aware Simulation (DAS) model, designed to incorporate dynamic attributes and enable data-aware decisions within simulations. The DAS model addresses the limitations of traditional approaches categorising attributes into 3 types, case, global, and event, to cover different scopes (local or global) and behaviour (static or dynamic) of the attributes. These attributes provide branching conditions at the decision points based on the current data state to guide the execution flow of the process.
Another significant aspect of this research is the discovery of the DAS model from the event logs. By incorporating the DAS model with discovery tools, organisations can discover their simulation models, including the perspective of data attributes and branching conditions that often affect the execution flow of business processes. Following discovery, organisations can simulate the model and make necessary adjustments and optimisations to adapt the models to reflect changes in operational procedures or to explore different ’what-if’ scenarios, thereby maintaining their relevance and effectiveness in dynamic business environments.
The evaluation demonstrates that data-aware models, discovered from event logs, can accurately classify data attributes, their update mechanisms, and implications into branching conditions. These models also replicate the control flow of the original log while enhancing cycle and event times, in contrast to traditional non-data-aware models that depend on branching probabilities.
Graduation Thesis language
English
Graduation Thesis type
Master - Software Engineering
Supervisor(s)
Orlenys López-Pintado, Marlon Dumas
Defence year
2024