1. Block-Structured Process Discovery: Filtering Infrequent Behaviour Sander Leemans Dirk Fahland Wil van der Aalst Eindhoven University of Technology

2. Process discovery Sander Leemans2 Fast Fitting Precise General Simple Sound

3. Trade-off ILPα Heuristics MinerEvolutionary Tree Miner not fast not fitting not sound not fitting not sound not simple not sound 3 Flower model not precise

4. Infrequent behaviour Sander Leemans4 80% model Filtering beforehand is difficult Filter during discovery

5. Outline Sander Leemans5 ?

6. Process trees Sander Leemans ab cd e a b cd e 6

7. Outline Sander Leemans7 ?

8. Divide & conquer {,, } {,, } {,, } recurse a Sander Leemans8

9. Finding operator {,, } {,, } recurse a Sander Leemans9 Find cut in directly-follows graph Sequence: edges crossing one- way only bc d e a

10. Inductive Miner Divide activities, select operator Split log Recurse until base case Sander Leemans ? ? {c,d} {a,b} {c}{d} 10

11. Outline Sander Leemans11 ?

12. Inductive Miner - infrequent Divide activities, select operator Split log Recurse until base case Sander Leemans ? ? {c,d} {a,b} {c}{d} 12 Threshold

13. Divide activities, select operator Filter infrequent edges (b,c) 100 (b,d) 100 (b,e) 100 (b,a) 1 Sander Leemans13 bc d e a 1 100

14. 2 5 5 4 2 2 2 2 2 5 Divide activities, select operator Weaker log Use eventually-follows relation instead Amplifies “correct” edges Sander Leemans14 bc d e a 1 1 1 1 1 1 1 1 1 1 1 abc (b,c) 1 (b,d) 1 (b,e) 1 (b,a) 1

15. Split log Sander Leemans15 {a} {b}

16. Outline Sander Leemans16 ?

17. Comparison 12 logs, 5 miners Discover (< 2 hours) (Convert to Petri net) Measure Inductive Miner Inductive Miner - infrequent Heuristics Miner Integer Linear Programming Miner Evolutionary Tree Miner (Flower model) (Trace model) Sander Leemans17

18. Comparison Sander Leemans18

19. Comparison Sander Leemans19

20. Comparison Sander Leemans20

21. Comparison Sander Leemans21

22. Comparison Sander Leemans22

23. Comparison Sander Leemans23

24. You have been watching In order of appearance Sander Leemans24 ? ?

25. Base cases Enough empty traces: filter Average number of events per trace close enough to 1 Sander Leemans25 1000 1 a 1000 a τ 1 1000 x τ 1