Presentation is loading. Please wait.

Presentation is loading. Please wait.

Understanding Deviations AIS Nov.2014 Elham Ramezani Dirk Fahland Wil M.P. van der Aalst.

Similar presentations


Presentation on theme: "Understanding Deviations AIS Nov.2014 Elham Ramezani Dirk Fahland Wil M.P. van der Aalst."— Presentation transcript:

1 Understanding Deviations AIS Nov.2014 Elham Ramezani Dirk Fahland Wil M.P. van der Aalst

2 Problem statement Event log 2 Rule 1 Rule n Compliance checking for R 1 Compliance checking for R n Compliance checker Compliance checking result of Rule 1 Compliance checking result of Rule n Is there a relation between different violations? Can we identify conditions that lead to a violation? ✗ ✓ ✓ ✓ ✗ ✗ Condition (I) Condition (II)

3 Problem statement Event log 3 Rule 1 Rule n Compliance checking for R 1 Compliance checking for R n Compliance checker Compliance checking result of Rule 1 Compliance checking result of Rule n Is there a relation between different violations? Can we identify conditions that lead to a violation?✓ ✗ ✓ ✓ ✗✓ Pattern discovery ✗ Can we discover distinguishable patterns in violating and compliant traces?

4 4 Collect all the compliance relate data Solution Overview

5 Event log 5 Rule 1 Rule n Compliance checking for R 1 Compliance checking for R n Compliance checker Compliance checking result of Rule 1 Compliance checking result of Rule n Compliance DB

6 6 Collect all the compliance relate data Solution Overview Connect the data obtained from different sources

7 Solution Overview Event log 7 Rule 1 Rule n Compliance checking result of Rule 1 Compliance checking result of Rule n Compliance DB Traces Events Attributes Rule description Rule type Analyzing attribute Alignments Moves Violating events Violation types Compliant value(s)

8 8 Collect all the compliance relate data Solution Overview Connect the data obtained from different sources Project all the diagnostics on the event log

9 Solution Overview 9 Compliance DB Enriched event log Traces Events Attributes Rule description Rule type Analyzing attribute Alignments Moves Violating events Violation types Compliant value(s) Enrich event log

10 10 Collect all the compliance relate data Solution Overview Connect the data obtained from different sources Project all the diagnostics on the event log Apply different machine learning and data mining technique

11 Solution Overview 11 Compliance DB Enriched event log Traces Events Attributes Rule description Rule type Analyzing attribute Alignments Moves Violating events Violation types Compliant value(s) Enrich event log Collection of data mining and machine learning techniques Patterns Relation between violations Condition lead to violations

12 12 activity: A activity: B Attribute x > 1000 … Resource 1 Resource 5 Resource 2 … Violations of Rule 1 Violations of Rule 2 Violations of Rule n Attribute y is ‘high’ 80% 50% …

13 13 Collect all the compliance relate data Solution Overview Connect the data obtained from different sources Project all the diagnostics on the event log Apply different machine learning and data mining technique Understand Deviations

14 Example 14 activity nameX-rayQuick-relief medication blood testdiagnosisControlling medication event IDe1e1 e2e2 e3e3 e4e4 e5e5 timet1t1 t2t2 t3t3 t4t4 t5t5 resourceSaraJohnSaraJohn trace ID01 History of heart attack yes log IDl1l1 l1l1 l1l1 l1l1 l1l1

15 Event log 15 Collecting all related compliance data Compliance DB event log related tables

16 Event log 16 Populating compliance DB with event log Compliance DB event log related tables Export and load event log to DB Sh.Shabani et al.

17 17 trace IDlog ID event IDtrace ID log ID log trace event attr. IDattr.keyattr.type attribute attr. IDevent IDvalue event-attribute-value attr. IDtrace IDvalue trace-attribute-value attr. IDlog IDvalue log-attribute-value activity nameX-rayquick-relief med.blood testdiagnosiscontrolling med. event IDe1e1 e2e2 e3e3 e4e4 e5e5 timet1t1 t2t2 t3t3 t4t4 t5t5 resourceSaraJohnSaraJohn trace ID01 heart attack statusyes log IDl1l1 l1l1 l1l1 l1l1 l1l1 time attr.01 e4e4 t4t4 01 l1l1 l1l1 TIME e4e4 attr.01 Populating compliance DB with event log

18 Event log 18 Populating compliance DB with event log Compliance DB event log related tables Export and load event log to DB Sh.Shabani et al.

19 19 Populating compliance DB with compliance rules info Compliance DB event log related tables Compliance rules info Compliance profile creator Rule 1 Rule n

20 20 compliance req. IDcompliance req. text compliance requirement compliance rule IDcompliance req. IDrule typerule textmeasure compliance rule Medical guideline diagnosis procedure temporal rule Rule 1time Req. 01 Diagnosis procedure must not be done later than time t 3. Populating compliance DB with compliance rules info Req. 01

21 21 Compliance profile creator Select compliance profile Compliance info object Event log Select compliance rule Compliance DB event log related tables Compliance rules info

22 22 Mapping a compliance rule to an event log trace IDlog ID event IDtrace ID trace event attr. IDattr.keyattr.type attribute attr. IDevent IDvalue event-attribute-value attr.01 e4e4 TIME time 01 l1l1 attr.01 e4e4 t4t4 compliance rule IDcompliance req. IDrule typerule textmeasure compliance rule temporal rule Rule 1timeReq.0 1 attr.IDmeasure attribute-measure time attr.01

23 23 Select compliance profile Compliance info object Event log Compliance checker Compliance checking result Export compliance checking result Compliance DB event log related tables Compliance rules info Compliance checking info

24 Example 24 Rule 1 : Diagnosis procedure must not be done later than time t 3. activity nameX-rayQuick-relief medication blood testdiagnosisControlling medication event IDe1e1 e2e2 e3e3 e4e4 e5e5 timet1t1 t2t2 t3t3 t4t4 t5t5 resourceSaraJohnSaraJohn trace ID01 History of heart attack yes diagnosis t4t4 ✗ log IDl1l1 l1l1 l1l1 l1l1 l1l1

25 25 Populating compliance DB with compliance checking info move ID Violation type Compliant value event ID alignment ID Compliance rule ID move Violation IDtype violation Compliance rule ID Trace ID alignment

26 26 Populating compliance DB with compliance checking info move ID Violation ID Compliant value event ID alignment ID move Violation IDtype violation Compliance rule ID trace ID alignment attr.IDmeasure attribute- measure compliance rule IDcompliance req. IDrule typerule textmeasure compliance rule event IDtrace ID event attr. IDattr.keyattr.type attribute e4e4 TIME time 01 temporal rule Rule 1Req.0 1time v.3 attr.01 time v.3 temporal violation m.4 t3t3 e4e4 01 Rule 1 01 Rule 1

27 27 Collect all the compliance relate data Solution Overview Connect the data obtained from different sources Project all the diagnostics on the event log Apply different machine learning and data mining technique Understand Deviations ✓ ✓

28 28 Enrich the event log with diagnostics Compliance DB Collect all diagnostics about a log Take all the moves related to a trace Generate events for model moves Project obtained diagnostics on events Create the enriched event log Enriched event log

29 29 Collect all the compliance relate data Solution Overview Connect the data obtained from different sources Project all the diagnostics on the event log Apply different machine learning and data mining technique Understand Deviations ✓ ✓ ✓

30 30 Root-cause analysis using process cube Compliance DB Collect all diagnostics about a log Enriched event log

31 31 Enriched event log Events enriched with diagnostics Root-cause analysis using OLAP

32 32 Enriched event log Root-cause analysis different data mining techniques Clustering Classification Association rule learning further diagnostics

33 33 Collect all the compliance relate data Solution Overview Connect the data obtained from different sources Project all the diagnostics on the event log Apply different machine learning and data mining technique Understand Deviations ✓ ✓ ✓ ✓

34 34 Thank you for your attention!


Download ppt "Understanding Deviations AIS Nov.2014 Elham Ramezani Dirk Fahland Wil M.P. van der Aalst."

Similar presentations


Ads by Google