Robust Process Mining with Guarantees: Process Discovery, Conformance Checking and Enhancement (Lecture Notes in Business Information Processing)

✍ Scribed by Sander J. J. Leemans

• Publisher: Springer
• Year: 2022
• Tongue: English
• Leaves: 478
• Category: Library

✦ Synopsis

This book presents techniques for process discovery, conformance checking and enhancement. For process discovery, it introduces the Inductive Miner framework: a recursive skeleton for discovery techniques that in itself provides several guarantees.

The framework is instantiated in several concrete discovery techniques, each of which targets a specific challenge of process discovery, such as incompleteness of information or noisy behavior. For conformance checking, it introduces the Projected Conformance Checking framework, which focuses on speed, but nevertheless provides several guarantees, such as that for certain classes of models, it can decide language equivalence. For enhancement, it introduces the Inductive visual Miner, a well-polished end-user focused tool that includes process discovery, conformance checking and that can visualize performance on a discovered model, all without any user input.

✦ Table of Contents

Preface
Acknowledgements
Summary
Contents
1 Introduction
1.1 Abstractions in Process Mining
1.2 Process Discovery
1.3 Conformance Checking
1.4 Enhancement & Tool Support
1.5 Contributions and Structure of this Book
References
2 Preliminaries
2.1 Multisets, Traces, Regular Expressions
2.2 Process Models
2.2.1 Automata
2.2.2 Petri Nets
2.2.3 Yet Another Workflow Language
2.2.4 Business Process Model and Notation
2.2.5 Process Trees
2.3 Event Logs
2.3.1 Atomic Event Logs
2.3.2 Non-Atomic Event Logs
2.3.3 Richer Logs
2.4 Directly Follows Relation
References
3 Process Mining
3.1 Different Use Cases, Different Process Mining Techniques
3.2 Formal Key Challenges of Process Mining
3.2.1 Models with Precise Semantics
3.2.2 System - Log - Model Relations
3.2.3 Simplicity & Balancing Log Criteria
3.2.4 An Ideal Technique (1)
3.3 Process Discovery
3.3.1 Discovery Algorithms Guaranteeing Soundness
3.3.2 Other Discovery Algorithms
3.3.3 An Ideal Process Discovery Technique (2)
3.4 Conformance Checking
3.4.1 Log Conformance Checking
3.4.2 System Conformance Checking
3.4.3 An Ideal Conformance Checking Technique (2)
3.5 Enhancement & Tool Support
3.5.1 Enhancements
3.5.2 Process Mining Tools
3.5.3 Requirements for Tool Support Beyond Process Discovery and Conformance Checking
3.6 Our Approach
3.6.1 A Process Discovery Framework
3.6.2 A Conformance Checking Framework
3.6.3 Enhancement & Tool Support
3.6.4 Future Work
References
4 Recursive Process Discovery
4.1 Recursive Process Discovery
4.1.1 An Example of Recursive Process Discovery
4.1.2 The IM framework
4.1.3 More Technical Examples
4.1.4 Guarantees
4.2 Rediscoverability
4.2.1 Rediscoverability using Abstractions
4.2.2 Rediscoverability and the IM framework
References
5 Abstractions
5.1 A Canonical Normal Form for Process Trees
5.1.1 Reduction Rules
5.1.2 Canonicity of the Reduction Rules
5.2 Language Uniqueness with Directly Follows Graphs
5.2.1 A Class of Trees: Cb
5.2.2 Footprints
5.2.3 Language Uniqueness
5.3 Language Uniqueness with Activity Relations
5.3.1 Activity Relations
5.3.2 Binary Trees
5.3.3 Language Uniqueness
5.4 Language Uniqueness with Interleaving
5.4.1 Footprint
5.4.2 A Class of Trees: Ci
5.4.3 Language Uniqueness
5.5 Language Uniqueness with Minimum Self-Distance
5.5.1 Minimum Self-Distance
5.5.2 A Class of Trees: Cm
5.5.3 Footprints
5.5.4 LC-Property
5.5.5 Language Uniqueness
5.6 Language Uniqueness with Optionality & Inclusive Choice
5.6.1 Optionality
5.6.2 Optionality in the Directly Follows Graph
5.6.3 A Class of Trees: Ccoo
5.6.4 Optionality under Sequence
5.6.5 Optionality under Inclusive Choice & Concurrency
5.6.6 Language Uniqueness
5.7 Language Uniqueness with non-Atomic Process Models
5.7.1 Non-Atomic Process Models
5.7.2 Representational Bias of Non-Atomic Models
5.7.3 Non-Atomic Directly Follows Graphs & Footprints
5.7.4 Concurrency Graphs & Footprints
5.7.5 A Class of Trees: Clc
5.7.6 Language Uniqueness
5.8 Classes of Process Trees: Revisited
References
6 Discovery Algorithms
6.1 Inductive Miner (IM)
6.1.1 Example
6.1.2 Inductive Miner (IM)
6.1.3 Guarantees
6.2 Handling Deviating & Infrequent Behaviour
6.2.1 Deviating & Infrequent Behaviour
6.2.2 Inductive Miner - infrequent (IMf)
6.2.3 Example
6.2.4 Guarantees
6.3 Handling Incomplete Behaviour
6.3.1 Incomplete Behaviour
6.3.2 Inductive Miner - incompleteness (IMc)
6.3.3 Example
6.3.4 Guarantees
6.3.5 Finding Cuts: Translation to SMT
6.4 Handling More Constructs: , 3942"613A``4547"603A and3942"613A`4547`"603A
6.4.1 Example
6.4.2 Inductive Miner - all operators (IMa)
6.4.3 Inductive Miner - infrequent - all operators (IMfa)
6.4.4 Guarantees
6.5 Handling Non-Atomic Event Logs
6.5.1 Non-Atomic Event Logs
6.5.2 Inductive Miner - life cycle (IMlc)
6.5.3 Inductive Miner - infrequent - life cycle (IMflc) & Inductive Miner - incompleteness - life cycle (IMclc)
6.5.4 Implementation
6.5.5 Guarantees
6.6 Handling Large Event Logs
6.6.1 Example
6.6.2 Inductive Miner - directly follows based framework (IMd framework)
6.6.3 Inductive Miner - directly follows (IMd)
6.6.4 Inductive Miner - infrequent - directly follows (IMfd)
6.6.5 Inductive Miner - incompleteness - directly follows (IMcd)
6.6.6 Guarantees
6.7 Tool Support
6.8 Summary: Choosing a Miner
References
7 Conformance Checking
7.1 Projected Conformance Checking Framework
7.1.1 Log to Projected Log to DFA
7.1.2 Model to Projected Model to DFA
7.1.3 Comparing DFAs & Measuring
7.1.4 Measuring over All Activities
7.2 An Example of Non-Conformance and Diagnostic Information
7.3 Guarantees
7.4 Tool Support
7.5 Conclusion
7.6 Ideas to Handle Unbounded & Weakly Unsound Petri Nets
References
8 Evaluation
8.1 Evaluated Process Discovery Algorithms
8.2 Scalability of Discovery Algorithms
8.2.1 Set-up
8.2.2 Results
8.2.3 Discussion
8.3 Log-Quality Dimensions
8.3.1 Event Logs
8.3.2 Quantitative
8.3.3 Qualitative
8.3.4 Conclusion
8.4 Rediscoverability & its Challenges
8.4.1 Incomplete Behaviour
8.4.2 Deviating & Infrequent Behaviour
8.5 Evaluation of Log-Conformance Checking
8.5.1 Set-up
8.5.2 Results
8.5.3 Discussion
8.5.4 Evaluation Using the PCC framework
8.6 Non-Atomic Behaviour
8.6.1 Artificial Log
8.6.2 Real-Life Log
8.7 Conclusion
References
9 Enhancement & Inductive visual Miner
9.1 Inductive visual Miner (IvM)
9.1.1 Steps & Architecture
9.1.2 Model Visualisation
9.1.3 Controls & Parameters
9.1.4 Adding Extensions
9.2 Deviations
9.2.1 Deviations and the PCC framework
9.2.2 Deviations and Alignments
9.3 Frequency Information
9.4 Projecting Performance Information on Process Trees
9.5 Animation
9.6 Conclusion
References
10 Conclusion
10.1 Process Discovery
10.2 Conformance Checking
10.3 Enhancement & Tool Support
10.4 Remaining Challenges
10.4.1 Detailed
10.4.2 Future Work
References
Index