Time Molecules: The Business Intelligence Side of Process Mining and Systems Thinking, by Eugene Asahara
Time is not just a measure—it’s the key to unlocking the hidden patterns that drive every system, from business intelligence to AI-powered decision-making.
Playbook to Reality: DNA and System Design
RNA: The Builder of Systems
Peptides to Proteins: Events and Markov Models
Peptides to Proteins: From Logs to Models
The Living System of Business Intelligence
Evolution
Models of How the World Works
Cycles
Why a Book Founded on Markov Models?
A Quick and Simple Example
Throw the Ball to Where the Receiver Will Be
The TL;DR of the Book for the BI Data Engineer
A Few Notes About this Book
Time Molecules vs TimeSolution vs Markov Models
Vendor Agnostic
Sequel to Enterprise Intelligence
Minimum Math and Code
Supplementary Material
GitHub Repository
My Blog Site
Intent of Supplementary Materials
Systems Thinking
The Mysteries of Natural Systems
Systems Evolving in a Chaotic World
Designed Systems vs Emergent Systems
Why Systems Thinking Matters for Time Molecules
Process Mining: Bridging the Gap Between Theory and Reality
The Challenge of Misalignment
Why Process Mining Matters
Systems Thinking in Process Mining
Process Mining in Today’s Business Environment
Leaks and Friction
Building Hypotheses and Markov Models
Why This Book Doesn’t Cover Petri Nets
Business Intelligence
Slicing and Dicing of Sets of Tuples
Applying These Concepts in Time Molecules
Time Molecules are Another Fact Aggregation
Time Molecules are Like OLAP Aggregations
At-Scale Analysis and the Role of Time Molecules
On-the-Fly Computation of Time Molecules
Digital Twins: The Other Side of the Process Mining Coin
A Non-Traditional Perspective: Nodes as Events, Not States
Events and Processes
The Markov Property
The Value of Heuristics and “System 1”
Transition Matrix
Stationary Distribution
Markov Model Manifold
Strategy: The Systems Engineering of Competing Goals
Time Series and Sequences
Video are “Time Pictures”
Markov Models In Machine Learning and AI
The TL;DR as a Large Language Model Prompt
The Time Molecules Warehouse
Bridging Process and Insight
Prep Work Before Time Molecules
Process Mining
Event Storming: Process Mining’s 2nd Cousin
Master Data Management (MDM)
How MDM and Process Mining Are Counterparts
Semantic Web and LLM Integration
Retrieval-Augmented Generation as an Orchestrator
Exploring Simple Sequences with Time Molecules
High-Level Architecture
Internet of Things
Event Logs and Streaming Data
Events and Subprocesses
Azure Event Hub: Real-Time Event Streaming
Set up SQL Server
Play Along at Home (or not)
Alternatives to SQL Server
At Scale Deployment Considerations
Markov Models on GPUs
TimeSolution Database
Restore TimeSolution Database
Event and Markov Models Ensembles
The Bronze, Silver, and Gold Medallion Framework for TimeSolution
Event Ensemble
Markov Model Ensemble
SelectedEvents TVF
MarkovProcess TVF
DimDate and DimTime
DimEvents Table
Window Functions in Event Streaming
IsState Flag
Synthetic Events from Machine Learning
Sources Table – Data Sources Dimension
DimObservers Table – Observers
EventSets Table
Transforms and Abstraction
Vectorized Transforms for Aggregation
Cases
Categorizing Characteristics of Events
Time Crystals
Perfect Cycles
Cases Table
CaseTypes Table
Case Properties Table
EventsFact Table
Event Metrics
Time Between Events – The Default Metric
Metrics Table
Metric Statistics
Open Schema Properties
Event Properties
Open Schema Properties to an Outrigger Dimension Table
Models Table
ModelEvents Table
Segment Columns
Transition Analysis Metrics Columns
Sequence and Ordinal Statistics Columns
ModelSequences Table
Function Example
LLMs and Code Generation
Procuring Events
AdventureWorksDW Example
Event Granularity
Event as an Effect
Date Best Practices for Events
Create Sample Markov Objects
Create DimEvents
Parse Case and Event Properties
Create Event Sets – InsertEventSets Stored Procedure
IsSequence
EventSetKey
Transforms
Create Markov Process
Model Similarity
Fraud Detection in Fintech
Car Dealership Sales Process
Key Metrics for Comparison
Setting Up Similarities
Model Stationary Distribution
Markov Model Utilization
Basic Markov Model
Fuel Metric as Opposed to Time Between
Compare Two Markov Models
Markov with No Transforms
Markov with Transforms
Caching Markov Models into the Markov Model Ensemble
Create a Code for the Transform
Cache the Markov Model with Transforms
Create a Focused Model
Find Models by Selected Properties
Find Models by Different Metric
Model Changes Across Dimensions
Model Drillthrough
Lateral Intersegment Event Scan
Rare Events
Model Event Anomalies – The EventPairAnomalies Table
Enumerate Multiple_Event Occurrences
Probabilities for all Sequences
Predict Next Event
Advanced Time Molecules Analytics Topics
Higher Order Markov Models
Markov Models of Time-Based Cases
Markov Model Confidence and Support
Support
Confidence
Additive Aspects of Markov Models
Non-Additive Metrics of Markov Models
Asynchronous Markov Model Creation
BayesianProbabilities Table
Generate Prompt for LLM Ratings of Events
Generate LLM Prompts for Generating Object Descriptions
Get Semantic Web and LLM Descriptions
Encoding System Enterprise Data-Driven Information into LLMs
Inductive Reasoning – Correlations
Markov Models – Inductive by Nature
Bayesian and Conditional Probabilities – A Hybrid Approach
Hidden Markov Models – The Deductive Leap
Piecing It All Together: A Spectrum of Reasoning
Let’s Try that Again with Another Example
Deductive Reasoning with Prolog
Knowledge Graph Beyond Ontologies and Taxonomies
Set up the Environment
Insight Space Graph Markov Models
Tuple Correlation Web
Conditional Probabilities in the TCW
Custom Correlation Scores
Time Molecules and the Data Catalog
Leveraging the Knowledge Graph
Drilling Up for Hierarchical Aggregation
Semantic Context for Transforms
Linking Markov Models Across Domains with the Knowledge Graph
Case Entry and Exit Points
Open Events Onboarding
Discovering Processes
Event Sets to Parse Cases
Parallel Sub-Processes
MMs are Like Hypotheses and NFAs are Like Theories
Workflow Machines
Probabilities to Workflow: Markov to NFA
From Markov to NFA
NFAs in Event Processing
Pick Two
Situation, Background, Analysis, Recommendation – SBAR
Situation: Defining the Problem
Background: What Has Changed?
Analysis: Finding Cause and Effect
Leveraging MMs for Process Dynamics
Bayesian/Conditional Reasoning for Cause and Effect
Iterative Hypothesis Testing
Recommendation: A Diagnosis
The Deliverable: Strategy as a Plan
Crafting the Solution Strategy
Delivering the Plan
Conclusion to the Artificial Consultant to the Consultant
In Time Molecules, Eugene Asahara introduces a business intelligence-inspired extension to process mining and systems thinking, blending Markov models, event logs, and AI-driven analytics to further enhance the cooperation of our human intelligence with enterprise intelligence. This book challenges traditional BI methods by shifting the focus from static snapshots of data to the fluid, interconnected nature of processes, empowering decision-makers to anticipate change, optimize workflows, and drive efficiency at scale.
Built on decades of experience in OLAP cubes, dimensional modeling, and data-driven strategy, Time Molecules provides BI professionals, data engineers, and executives with the tools to model and analyze event-driven processes using a scalable, probabilistic framework. By treating time as the central dimension and events as molecular building blocks, Asahara demonstrates how Markov models can be applied at scale, offering a powerful method to transform raw data into actionable business intelligence.
Readers will explore how businesses, like living organisms, thrive through interconnected systems of processes, evolving dynamically in response to market forces, competition, and operational inefficiencies. Through process mining, digital twins, and AI-powered insights, Time Molecules reveals how organizations can uncover inefficiencies, optimize decision-making, and build resilience in an ever-changing world.
The book covers a wide range of topics essential for modern data professionals, including the integration of Markov models with traditional BI, leveraging the astonishing versatility of AI and large language models (LLMs), and applying systems thinking to complex business operations. From case studies in logistics, sales, and customer behavior to real-world implementations using SQL-based TimeSolution architecture, Time Molecules bridges theory and practice in an accessible and insightful manner.
With an emphasis on real-time analytics and predictive modeling, this book highlights how businesses can move beyond simple metrics like sums and counts to more sophisticated probabilistic forecasting. It explores how slicing and dicing event sequences across multiple dimensions provides a deeper understanding of customer journeys, operational bottlenecks, and hidden correlations in large-scale data environments.
Time Molecules also delves into the impact of AI on business intelligence, illustrating how machine learning models can enhance process mining and decision automation. It discusses the limitations of large language models (LLMs) in business operations and presents Markov models as a transparent and efficient alternative for time-based data analysis.
Designed for data professionals seeking to elevate their BI capabilities, this book serves as both a practical guide and a conceptual roadmap for mastering the art of process-driven intelligence. Whether you are a BI engineer, a CDO, or an executive navigating digital transformation, Time Molecules offers a structured approach to understanding the complexities of modern business processes.
Extending process mining and systems thinking with BI concepts, Asahara introduces a paradigm shift in how we perceive and analyze the flow of business events. The book provides practical insights into bridging the gap between theory and application, helping businesses harness the power of data to drive smarter, faster, and more adaptive decision-making.
With AI and automation redefining industries at an astounding and accelerating rate, Time Molecules is a must-read for anyone looking to future-proof their business intelligence strategy.
Eugene Asahara, with a rich history of over 45 years in software development, including over 25 years focused on business intelligence, particularly SQL Server Analysis Services (SSAS), is currently working as a Principal Solutions Architect at Kyvos Insights. His exploration of knowledge graphs began in 2005 when he developed Soft-Coded Logic (SCL), a .NET Prolog interpreter designed to modernize Prolog for a data-distributed world. Later in 2012, Eugene ventured into creating Map Rock, a project aimed at constructing knowledge graphs that merge human and machine intelligence across numerous SSAS cubes. While these initiatives didn’t gain extensive adoption at the time, the lessons learned have proven invaluable. With the emergence of Large Language Models (LLMs), building and maintaining knowledge graphs has become practically achievable, and Eugene is leveraging his past experience and insights from SCL and Map Rock to this end. That experience is encapsulated in his books, Enterprise Intelligence and Time Molecules.
Please complete all fields.
