RICHARD MÜLLER WIL VAN DER AALST CHRISTIAN STAHL

Transcription

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