Process Mining ControlFlow Mining Algorithms Ana Karla Alves

Process Mining: Control-Flow Mining Algorithms Ana Karla Alves de Medeiros Eindhoven University of Technology Department of Information Systems a. k. medeiros@tue. nl /faculteit technologie management 1

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 2

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 3

Process Mining Event Logs 1. 2. 3. 4. 5. 6. 7. /faculteit technologie management 1. Register at hospital 2. Talk to 1. doctor Register at hospital 3. Undergo surgery hospital 1. Register at hospital Talk 4. to Take doctor medicines to doctor 2. 2. Talk to doctor 1. Register at. Talk Take 5. blood test Take blood sample hospital Undergo surgery 3. 3. Undergo surgery Take 6. X-ray 1. to Register at hospital Talk doctor 2. Talk to doctor 4. Take medicines Get 7. results 2. Talk to doctor Go 5. home 3. Undergo surgery 5. Take blood sample Talk to doctor 3. Undergo surgery 4. Take 6. medicines 6. Talk to doctor Go home 4. Take medicines 5. Take 7. blood sample 7. home Go. Go home 5. doctor Take blood sample 6. Talk to doctor 7. Go home 4

Process Mining Process Model Organizational Model Social Network Performance Analysis Auditing/Security Event Log Mining Techniques /faculteit technologie management Mined Models 5

Event Logs are Everywhere! aaaaaa aaaaaa Machines, Municipalities, Airports, Internet, Hospitals, etc. /faculteit technologie management 6

Tools • • www. processmining. org Pro. M 4. 2 Pro. Mimport Free tools! /faculteit technologie management 7

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 8

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 9

Types of Algorithms /faculteit technologie management 10

Types of Algorithms Process Model Organizational Model Social Network /faculteit technologie management 11

Types of Algorithms Compliance Process Model Auditing/Security /faculteit technologie management 12

Types of Algorithms Bottlenecks/ Business Rules Process Model Performance Analysis /faculteit technologie management 13

Process Mining Start Get Ready 1. 2. 3. 1. 4. 2. 5. 3. 6. 4. 7. 5. 8. 6. 7. 8. 9. 10. Start 1. Start Get Ready 1. 2. Start Ready Travel by. Get Train 2. Start Get Ready 3. Travel by Train Beta Event Starts 3. Get Travel by Car Ready 4. Beta Event Starts Visit Brewery 4. Travel Beta Event Starts by Car 5. Give a Talk Have Dinner 5. Conference Give a. Starts Talk 6. Brewery Go Home. Visit 6. Join Visit. Have Brewery Reception 7. Travel by Train. Dinner 7. Have 8. Dinner Go. Dinner Home 8. Go Go Home 9. Travel by Train 9. Pay Parking 10. Travel by Car End Travel by Train Travel by Car Conference Starts Give a Talk Join Reception Mined Model Have Dinner Go Home Event Log Pay Parking Travel by Train Discovery Techniques: Control-Flow Mining /faculteit technologie management Travel by Car End 14

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 15

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 16

Common Constructs • • Sequence Splits Joins Loops Non-Free Choice Invisible Tasks Duplicate Tasks Get Ready Travel by Train Travel by Car Defense Starts Ask Question Give a Talk Defense Ends Have Drinks Go Home Travel by Train Pay Parking Travel by Car /faculteit technologie management 17

Common Constructs • • Sequence Splits Joins Loops Non-Free Choice Invisible Tasks Duplicate Tasks + noise in logs! /faculteit technologie management Get Ready Travel by Train Travel by Car Defense Starts Ask Question Give a Talk Defense Ends Have Drinks Go Home Travel by Train Pay Parking Travel by Car 18

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 19

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 20

Event Log: Mining XML (MXML) The notion of which tasks belong to a same instance is crucial for applying process mining techniques! /faculteit technologie management 21

Event Log: Mining XML (MXML) /faculteit technologie management 22

Event Log: Mining XML (MXML) Compulsory fields! /faculteit technologie management 23

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 24

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 25

α-algorithm 1. 2. 3. 4. 5. Read a log Get the set of tasks Infer the ordering relations Core Step! Build the net based on inferred relations Output the net /faculteit technologie management 26

α-algorithm - Ordering Relations >, , ||, # • Direct succession: x>y iff for some case x is directly followed by y. • Causality: x y iff x>y and not y>x. • Parallel: x||y iff x>y and y>x • Unrelated: x#y iff not x>y and not y>x. /faculteit technologie management case case case. . . 1 2 3 3 1 1 2 4 2 : : : : : task task task A A A B B C C A B A>C B>D C>B C>D E>F ABCD ACBD EF B||C C||B A C B D C D E F 27

α-algorithm - Formalization Let W be a workflow log over T. a(W) is defined as follows. 1. TW = { t Î T | $s Î W t Î s}, 2. TI = { t Î T | $s Î W t = first(s) }, 3. TO = { t Î T | $s Î W t = last(s) }, 4. XW = { (A, B) | A Í TW Ù B Í TW Ù "a Î A"b Î B a W b Ù "a 1, a 2 Î A a 1#W a 2 Ù "b 1, b 2 Î B b 1#W b 2 }, 5. YW = { (A, B) Î X | "(A¢, B¢) Î XA Í A¢ ÙB Í B¢Þ (A, B) = (A¢, B¢) }, 6. PW = { p(A, B) | (A, B) Î YW } È{i. W, o. W}, 7. FW = { (a, p(A, B)) | (A, B) Î YW Ù a Î A } È { (p(A, B), b) | (A, B) Î YW Ù b Î B } È{ (i. W, t) | t Î TI} È{ (t, o. W) | t Î TO}, and 8. a(W) = (PW, TW, FW). /faculteit technologie management 28

ABCD ACBD EF B||C C||B /faculteit technologie management A B A C B D C D E F α-algorithm - Insight 29

α-algorithm – Log properties + target nets • If log is complete with respect to relation >, it can be used to mine SWF-net without short loops • Structured Workflow Nets (SWF-nets) have no implicit places and the following two constructs cannot be used: /faculteit technologie management 30

α-algorithm – No short loops B>B and not B>B implies B B (impossible!) Why no short loops? One-length Two-length A>B and B>A implies A||B and B||A instead of A B and B A /faculteit technologie management 31

α-algorithm – Common Constructs Why no duplicate tasks? Why not invible tasks? Why noisefree logs? • No invisible tasks, non-free-choice or duplicate tasks • No noisy logs /faculteit technologie management 32

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Algorithm Fuzzy Miner /faculteit technologie management 33

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Algorithm Fuzzy Miner /faculteit technologie management 34

Heuristics Miner 1. Read a log 2. Get the set of tasks 3. Infer the ordering relations based on their frequencies 4. Build the net based on inferred relations 5. Output the net /faculteit technologie management 35

Heuristics Miner Insight: The more frequently a task A directly follows another task B, and the less frequently the opposite occurs, the higher the probability that A causally follows B! /faculteit technologie management 36

α-algorithm – Common Constructs Why no non-freechoice? Why no guarantee whole net will be correct? • No non-free-choice or duplicate tasks • Robust to invisible tasks and noisy logs /faculteit technologie management 37

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 38

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 39

Genetic Process Mining (GPM) • Genetic Algorithms + Process Mining • Genetic Algorithms – Search technique that mimics the process of evolution in biological systems • Advantages – Tackle all common structural constructs – Robust to noise • Disadvantages – Computational Time /faculteit technologie management 40

Genetic Process Mining (GPM) Algorithm: Internal Representation Fitness Measure Genetic Operators /faculteit technologie management 41

GPM – Fitness Measure Start • Guides the search! Get Ready Travel by Train Conference Starts Travel by Car Give a Talk Visit Brewery Have Dinner Go Home Pay Parking Travel by Train Travel by Car End /faculteit technologie management 42

GPM – Fitness Measure /faculteit technologie management 43

GPM – Fitness Measure Overgeneral solution Start Travel by Train Conference Starts Get Ready Give a Talk Have Dinner Visit Brewery Punish for the amount of Travel by Car enabled tasks during the parsing! /faculteit technologie management End Go Home Pay Parking 44

GPM – Fitness Measure Overspecific solution Start Get Ready Travel by Train Travel by Car Conference Starts Give a Talk Visit Brewery Have Dinner Go Home Punish for the amount of Travel by Train duplicate tasks with common input/output tasks! /faculteit technologie management Pay Parking Travel by Car End 45

GPM – DGA Pro. M Plug-in Why does the GA Miner takes so much time? How could we improve its running time without changing the code? /faculteit technologie management 46

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 47

Process Mining • • Short Recap Types of Process Mining Algorithms Common Constructs Input Format α-algorithm Heuristics Miner Genetic Miner Fuzzy Miner /faculteit technologie management 48

Fuzzy Miner - Motivation Mine less structured processes! /faculteit technologie management 49

Fuzzy Miner - Motivation /faculteit technologie management 50

Fuzzy Miner /faculteit technologie management 51

No “Ask Question” or Fuzzy Miner “Give Talk”! Abstracting even more from details! All details! /faculteit technologie management 52

Conclusions • • The notion of a process instance is crucial! Ordering of tasks is the basic information Frequencies are important to handle noise Local approaches – α-algorithm, Heuristics Miner • Global approaches – Genetic Miner and Fuzzy Miner /faculteit technologie management 53