[Paper Review] Advances in Process Optimization: A Comprehensive Survey of Process Mining, Predictive Process Monitoring, and Process-Aware Recommender Systems

Anzalee Khan, Aditya Ghose+2 more|arXiv (Cornell University)|Jan 25, 2023

Big Data and Business IntelligenceBusiness, Management and Accounting3 citations

TL;DR

This survey presents a comprehensive overview of process-oriented data science, focusing on process mining, predictive process monitoring, and process-aware recommender systems to support business process management. It integrates historical logs, real-time data, and machine learning to enable predictive insights, diagnostic analysis, and prescriptive recommendations for process optimization and automation.

ABSTRACT

Process analytics approaches allow organizations to support the practice of Business Process Management and continuous improvement by leveraging all process-related data to extract knowledge, improve process performance and support decision-making across the organization. Process execution data once collected will contain hidden insights and actionable knowledge that are of considerable business value enabling firms to take a data-driven approach for identifying performance bottlenecks, reducing costs, extracting insights and optimizing the utilization of available resources. Understanding the properties of 'current deployed process' (whose execution trace is often available in these logs), is critical to understanding the variation across the process instances, root-causes of inefficiencies and determining the areas for investing improvement efforts. In this survey, we discuss various methods that allow organizations to understand the behaviour of their processes, monitor currently running process instances, predict the future behavior of those instances and provide better support for operational decision-making across the organization.

Motivation & Objective

• To provide a holistic overview of process-oriented data science techniques that support business process management (BPM) and continuous improvement.
• To identify and analyze underexplored areas in process analytics, particularly in predictive monitoring and decision support.
• To bridge the gap between traditional process mining and advanced analytics by integrating diverse data sources and intelligent decision-making capabilities.
• To highlight the role of automation, including RPA and search-based optimization, in enabling proactive process improvement.
• To position process analytics as a strategic organizational capability for enhancing performance, reducing costs, and enabling data-driven decision-making.

Proposed method

• Utilizes process mining techniques such as process discovery, conformance checking, and enhancement to reverse-engineer process models from event logs.
• Applies predictive analytics to forecast process outcomes using historical and real-time process data, including cycle time, cost, and case completion probabilities.
• Employs recommendation systems to suggest optimal actions during process execution, such as task reordering, resource allocation, and decision rule optimization.
• Integrates robotic process automation (RPA) to automate repetitive, rule-based interactions between users and software systems.
• Leverages search-based optimization and machine learning (including causal inference) to identify improvement opportunities based on KPIs like cycle time and cost.
• Combines domain knowledge with historical process data to generate prescriptive recommendations for process redesign and real-time operational decisions.

Experimental results

Research questions

• RQ1 How can process analytics techniques extract actionable insights from process execution logs to support process improvement and decision-making?
• RQ2 What are the key capabilities of predictive process monitoring in forecasting process outcomes and identifying risks during execution?
• RQ3 How can process-aware recommender systems enhance operational decision-making in dynamic, knowledge-intensive processes?
• RQ4 In what ways can automated process improvement, including RPA and optimization techniques, proactively enhance process performance?
• RQ5 How do integrated data sources (e.g., event logs, decision logs, provisioning logs) contribute to more comprehensive process understanding and optimization?

Key findings

• Process analytics extends beyond traditional process mining by incorporating diverse data types—such as decision logs and context data—to deliver predictive, diagnostic, and prescriptive insights.
• Predictive process monitoring enables early detection of potential delays or failures in running process instances, supporting timely managerial interventions.
• Process-aware recommender systems can improve process outcomes by suggesting optimal task sequences, resource allocations, or decision rules based on historical best practices.
• Automated process improvement using search-based optimization and machine learning can identify opportunities to reduce cycle time, lower costs, and enhance resource utilization.
• RPA integration enables automation of repetitive, user-facing tasks in business processes, especially where traditional automation is not cost-effective.
• Causal machine learning models applied during process execution can improve outcomes by recommending evidence-based treatment or routing decisions.