Private View Conformance Checking

Conformance checking techniques can be used to diagnose differences between observed behavior and modeled behavior. Although these techniques can be used to measure the degree of conformance of a running service based on recorded event data (e.g., messages or transaction logs) and its specification, their application may produce “false negatives” because a private view (i.e., an implementation) that accords with its specification may deviate significantly. The implementation may reorder some activities without introducing any problems, yet traditional conformance checking would penalize such changes unjustifiably. To overcome this problem, we present a novel approach that determines a best matching private view. We show that among the infinitely many accordant private views, there is a canonical best matching private view. Although the current implementation and experiments are limited to acyclic service models, the approach can also be applied to cyclic service models.

I gave this talk at the 9th International Workshop on Web Services and Formal Methods, WS-FM 2012, September 6-7, 2012, Tallinn, Estonia.

Transcript

1. PRIVATE VIEW CONFORMANCE CHECKING RICHARD MÜLLER WIL VAN DER AALST CHRISTIAN STAHL
2. CONTRACT ALICE BOB CONTRACT PARTIES MODEL A SERVICE-ORIENTED SYSTEM 1
3. PUBLIC VIEW 1 ALICE BOB
4. PUBLIC VIEW 1 ALICE BOB
5. PUBLIC VIEW 1 ASSUMPTION COMPOSITION OF PUBLIC VIEWS IS CORRECT ALICE BOB
6. IMPLEMENTATION 2 PUBB IMPLB PUBA IMPLA
7. IMPLEMENTATION 2 PUBB IMPLB PUBA IMPLA GOAL COMPOSITION OF IMPLEMENTATIONS IS CORRECT
8. ACCORDANCE RELATION 3 PUBB IMPLB PUBA IMPLA
9. ACCORDANCE RELATION 3 PUBB IMPLB PUBA IMPLA
10. ACCORDANCE RELATION 3 PUBB IMPLB PUBA IMPLA THEOREM IF IMPLX ACCORDS WITH PUBX FOR ALL PARTIES X, THEN COMPOSITION OF ALL IMPLX IS CORRECT
11. PRIVATE VIEW 3 PUBB IMPLB PUBA IMPLA THEOREM IF IMPLX ACCORDS WITH PUBX FOR ALL PARTIES X, THEN COMPOSITION OF ALL IMPLX IS CORRECT
12. PRIVATE VIEW 3 PUBB PUBA NOTATION A PRIVATE VIEW IS AN ACCORDANT IMPLEMENTATION PRIVB PRIVA
13. VIEWPOINT OF ALICE 4 PUBB PRIVB
14. VIEWPOINT OF ALICE 4 PUBB IMPLB ?
15. VIEWPOINT OF ALICE 4 PUBB IMPLB ?
16. VIEWPOINT OF ALICE 4 PUBB IMPLB ? LOGB EVENT LOG RECORDED BEHAVIOR OF IMPLEMENTATION
17. PROBLEM STATEMENT 4 PUBB IMPLB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH COULD HAVE PRODUCED THIS LOG?” ?
18. OUTLINE 4 PUBB IMPLB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH COULD HAVE PRODUCED THIS LOG?” ? 1. MODEL VS. EVENT LOG
19. OUTLINE 4 PUBB IMPLB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH COULD HAVE PRODUCED THIS LOG?” ? 1. MODEL VS. EVENT LOG 2. SKETCH SOLUTION
20. OUTLINE 4 PUBB IMPLB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH COULD HAVE PRODUCED THIS LOG?” ? 1. MODEL VS. EVENT LOG 2. SKETCH SOLUTION 3. PRESENT EXPERIMENTAL RESULTS
21. MODEL VS. EVENT LOG 5 PUBB IMPLB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH COULD HAVE PRODUCED THIS LOG?” ?
22. MODEL VS. EVENT LOG 5 PUBB IMPLB LOGB CONFORMANCE CHECKING COMPARING MODEL AND EVENT LOG ?
23. REPLAY 6 MODEL EVENT LOG ad bc t3 c a t4 p3 p2 p0 t1 t0 p1 b d t2
24. REPLAY 6 MODEL EVENT LOG ad bc t3 c a t4 p3 p2 p0 t1 t0 p1 b d t2
25. REPLAY 6 MODEL EVENT LOG ad bc t3 c a t4 p3 p2 p0 t1 t0 p1 b d t2
26. REPLAY 6 MODEL EVENT LOG ad bc t3 c a t4 p3 p2 p0 t1 t0 p1 b d t2
27. REPLAY 6 MODEL EVENT LOG ad bc t3 c a t4 p3 p2 p0 t1 t0 p1 b d t2
28. REPLAY 6 MODEL EVENT LOG ad bc t3 c a t4 p3 p2 p0 t1 t0 p1 b d t2 ?
29. REPLAY 6 MODEL EVENT LOG ad bc t3 c a t4 p3 p2 p0 t1 t0 p1 b d t2 ?
30. MORE PRECISE PROBLEM STATEMENT 7 PUBB IMPLB LOGB ? “IS THERE ANY PRIVATE VIEW OF BOB WHICH CONFORMS TO THIS LOG?”
31. SET VIEW ON PROBLEM STATEMENT 8 PUBB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH CONFORMS TO THIS LOG?”
32. SET VIEW ON PROBLEM STATEMENT 8 PUBB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH CONFORMS TO THIS LOG?” PRIVATE VIEWS
33. SET VIEW ON PROBLEM STATEMENT 8 PUBB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH CONFORMS TO THIS LOG?” PRIVATE VIEWS CONFORMING MODELS
34. SET VIEW ON PROBLEM STATEMENT 8 PUBB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH CONFORMS TO THIS LOG?” PRIVATE VIEWS CONFORMING MODELS
35. SET VIEW ON PROBLEM STATEMENT 8 PUBB LOGB “IS THERE ANY PRIVATE VIEW OF BOB WHICH CONFORMS TO THIS LOG?” PRIVATE VIEWS CONFORMING MODELS INFINITELY MANY INFINITELY MANY
36. BEST MATCHING PRIVATE VIEW 9 PUBB LOGB PRIVATE VIEWS CONFORMING MODELS FINITE CHARACTERIZATION
37. BEST MATCHING PRIVATE VIEW 9 PUBB LOGB PRIVATE VIEWS CONFORMING MODELS FINITE CHARACTERIZATION MB
38. 1. CASE: MB CONFORMS TO LOGB 10 PUBB LOGB PRIVATE VIEWS CONFORMING MODELS MB
39. 11 PUBB LOGB PRIVATE VIEWS CONFORMING MODELS MB 2. CASE: MB DOES NOT CONFORM TO LOGB
40. 11 PUBB LOGB PRIVATE VIEWS CONFORMING MODELS MB 2. CASE: MB DOES NOT CONFORM TO LOGB THEOREM MB CONFORMS TO LOGB IFF THERE EXISTS A PRIVATE VIEW OF PUBB WHICH CONFORMS TO LOGB
41. NECESSARY CONDITION 12 PUBB IMPLB LOGB ? COROLLARY IF MB DOES NOT CONFORM TO LOGB THEN IMPLB DOES NOT ACCORD WITH PUBB
42. EVALUATION 13 PUBB MB COMPUTE MB USING WENDY, MAXIS, PNAPI
43. EVALUATION 13 PUBB MB COMPUTE MB USING WENDY, MAXIS, PNAPI WWW.SERVICE-TECHNOLOGY.ORG
44. EVALUATION 13 PUBB MB COMPUTE MB USING WENDY, MAXIS, PNAPI LOGB WWW.SERVICE-TECHNOLOGY.ORG MEASURE CONFORMANCE USING PROM WWW.PROMTOOLS.ORG
45. EVALUATION 13 PUBB MB COMPUTE MB USING WENDY, MAXIS, PNAPI LOGB WWW.SERVICE-TECHNOLOGY.ORG + + MEASURE CONFORMANCE USING PROM MEASURE CONFORMANCE USING PROM WWW.PROMTOOLS.ORG
46. PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB |P| |T| |Interface| ∅ TIME ∅ COST |P| |T| COMPUTE MB ∅ TIME ∅ COST 14 6 6 3.47ms 6.21 35 132 < 1s 0.34ms 0.20 16 8 8 3.31ms 7.53 41 190 < 1s 0.88ms 0.20 30 18 9 6.21ms 8.26 106 681 3s 1.41ms 0.19 38 32 10 7.10ms 4.89 32 168 < 1s 0.17ms 0.05 88 74 11 33.93ms 7.24 806 6060 102s 45.60ms 0.03 EXPERIMENTAL RESULTS ARTIFICIAL ACYCLIC PUBB, WEAK TERMINATION 14
47. PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB |P| |T| |Interface| ∅ TIME ∅ COST |P| |T| COMPUTE MB ∅ TIME ∅ COST 14 6 6 3.47ms 6.21 35 132 < 1s 0.34ms 0.20 16 8 8 3.31ms 7.53 41 190 < 1s 0.88ms 0.20 30 18 9 6.21ms 8.26 106 681 3s 1.41ms 0.19 38 32 10 7.10ms 4.89 32 168 < 1s 0.17ms 0.05 88 74 11 33.93ms 7.24 806 6060 102s 45.60ms 0.03 EXPERIMENTAL RESULTS EXPONENTIAL IN INTERFACE SIZE 14
48. EXPERIMENTAL RESULTS PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B PUBLIC VIEW OF PARTY B BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB BEST MATCHING PRIVATE VIEW MB |P| |T| |Interface| ∅ TIME ∅ COST |P| |T| COMPUTE MB ∅ TIME ∅ COST 14 6 6 3.47ms 6.21 35 132 < 1s 0.34ms 0.20 16 8 8 3.31ms 7.53 41 190 < 1s 0.88ms 0.20 30 18 9 6.21ms 8.26 106 681 3s 1.41ms 0.19 38 32 10 7.10ms 4.89 32 168 < 1s 0.17ms 0.05 88 74 11 33.93ms 7.24 806 6060 102s 45.60ms 0.03 ARTIFICIAL LOGB, 100 TRACES, ~600 EVENTS 14
49. FUTURE WORK + IMPLEMENTATION FOR CYCLIC PUBLIC VIEWS + EVALUATION WITH INDUSTRIAL MODELS + OTHER CORRECTNESS CRITERIA + OTHER CONFORMANCE DIMENSIONS TAKE HOME POINTS WITH BEST MATCHING PRIVATE VIEW MB IMPLEMENTATION FOR ACYCLIC PUBLIC VIEW AND WEAK TERMINATION 1 2 15 DECIDE IF THERE EXISTS A PRIVATE VIEW OF PUBB WHICH CONFORMS TO EVENT LOG LOGB
50. FUTURE WORK + IMPLEMENTATION FOR CYCLIC PUBLIC VIEWS + EVALUATION WITH INDUSTRIAL MODELS + OTHER CORRECTNESS CRITERIA + OTHER CONFORMANCE DIMENSIONS TAKE HOME POINTS WITH BEST MATCHING PRIVATE VIEW MB IMPLEMENTATION FOR ACYCLIC PUBLIC VIEW AND WEAK TERMINATION 1 2 15 DECIDE IF THERE EXISTS A PRIVATE VIEW OF PUBB WHICH CONFORMS TO EVENT LOG LOGB
51. FUTURE WORK + IMPLEMENTATION FOR CYCLIC PUBLIC VIEWS + EVALUATION WITH INDUSTRIAL MODELS + OTHER CORRECTNESS CRITERIA + OTHER CONFORMANCE DIMENSIONS TAKE HOME POINTS WITH BEST MATCHING PRIVATE VIEW MB IMPLEMENTATION FOR ACYCLIC PUBLIC VIEW AND WEAK TERMINATION 1 2 15 DECIDE IF THERE EXISTS A PRIVATE VIEW OF PUBB WHICH CONFORMS TO EVENT LOG LOGB
52. FUTURE WORK + IMPLEMENTATION FOR CYCLIC PUBLIC VIEWS + EVALUATION WITH INDUSTRIAL MODELS + OTHER CORRECTNESS CRITERIA + OTHER CONFORMANCE DIMENSIONS TAKE HOME POINTS WITH BEST MATCHING PRIVATE VIEW MB IMPLEMENTATION FOR ACYCLIC PUBLIC VIEW AND WEAK TERMINATION 1 2 15 DECIDE IF THERE EXISTS A PRIVATE VIEW OF PUBB WHICH CONFORMS TO EVENT LOG LOGB
53. TAKE HOME POINTS WITH BEST MATCHING PRIVATE VIEW MB IMPLEMENTATION FOR ACYCLIC PUBLIC VIEW AND WEAK TERMINATION 1 2 FUTURE WORK + IMPLEMENTATION FOR CYCLIC PUBLIC VIEWS + EVALUATION WITH INDUSTRIAL MODELS + OTHER CORRECTNESS CRITERIA + OTHER CONFORMANCE DIMENSIONS 15 DECIDE IF THERE EXISTS A PRIVATE VIEW OF PUBB WHICH CONFORMS TO EVENT LOG LOGB
54. richard.mueller@informatik.hu-berlin.de http://about.me/richardmueller RICHARD MÜLLER WIL VAN DER AALST CHRISTIAN STAHL THANK YOU!