Overview%20of%20Today

1
Overview of Todays Talks

• Provenance Data Structures
• Recording and Querying Provenance
• Break (30 minutes)
• Distribution and Scalability
• Security
• Methodology

2
Process Documentation Concepts and Data
Structures
by Simon Miles (sm_at_ecs.soton.ac.uk)
3
Process Documentation

• The provenance of x is the process that led to x.
• How do we find the provenance of x?
• Two steps
• Document the process in an application
• Retrieve the process that led to X from the
  documentation

4
World View

• Every application is made up of actors
• Every change that happens is an action by an
  actor
• Actors communicate by sending messages.
• Every action is triggered by a message
• The outputs of (messages sent by) an actor are
  caused by the inputs to (messages received by)
  the actor

5
Interaction in the Organ Transplant Process
Request healthcare record for patient PID1
Donor Data Collector
Electronic Healthcare Record System
6
Request Message Contents

• ltsoapenvelopegt
• ltsoapheadergtlt/soapheadergt
• ltsoapbodygt
• ltechrsrequestgt
• ltechrspatientgt PID1 lt/echrspatientgt
• lt/echrsrequestgt
• lt/soapbodygt
• lt/soapenvelopegt

7
Identifying a Message

• An interaction is a message being sent by one
  actor and received by another
• An interaction is identified by the following
• The source of the message (startpoint)
• The sink of the message (endpoint)
• A unique ID that distinguishes between
  interactions with the same source and sink

8
Interaction Key
Web Services Endpoint Reference
9
Asserting the Content of a Message

• In asserting the content of a message it has sent
  or received, an actor provides the following
• A unique id for the assertion
• The style in which the content is asserted
• The content of the message

10
Interaction P-Assertion
11
Request Message Contents

• ltsoapenvelopegt
• ltsoapheadergtlt/soapheadergt
• ltsoapbodygt
• ltechrsrequestgt
• ltechrspatientgt PID1 lt/echrspatientgt
• lt/echrsrequestgt
• lt/soapbodygt
• lt/soapenvelopegt

12
Styled P-Assertion

• ltpsinteractionPAssertiongt
• ltpslocalPAssertionIdgt1lt/pslocalPAssertionIdgt
• ltpsdocumentationStylegt
• http//www.pasoa.org/.../stylesAnonymisedPa
  tient
• lt/psdocumentationStylegt
• ltpscontentgt
• ltsoapenvelopegt
• ltsoapheadergtlt/soapheadergt
• ltsoapbodygt
• ltechrsrequestgt
• ltechrsanoymisedPatientgt x93ab4 lt/
  echrsanoymisedPatientgt
• lt/echrsrequestgt
• lt/soapbodygt
• lt/soapenvelopegt
• lt/pscontentgt
• lt/psinteractionPAssertiongt

13
Asserting the State of an Actor

• In asserting its state, an actor provides the
  following
• A description of part or all of its state
• Optionally, the style in which it is asserted
• A unique id for the assertion

14
Actor State P-Assertion
15
Example (Donor Data Collector)

• ltpsactorStatePAssertiongt
• ltpslocalPAssertionIdgt 2 lt/pslocalPAssertionIdgt
  
• ltpscontentgt
• ltdonordataHeldOngt
• ltechrspatientgt PID5 lt/echrspatientgt
• ltechrspatientgt PID16 lt/echrspatientgt
• lt/donordataHeldOngt
• lt/pscontentgt
• lt/psactorStatePAssertiongt

16
Data Within Assertions

• Application messages and actor states contain
  parts, e.g. arguments to operations
• These parts can have independent provenance

17
Another Interaction
Donor Data Collector
Testing Lab
Blood Analysis Results
18
Analysis Results Message Contents

• ltsoapenvelopegt
• ltsoapheadergtlt/soapheadergt
• ltsoapbodygt
• lttestlabanalysisResultsgt
• lttestlabsampleIDgt Sample3
  lt/testlabsampleIDgt
• lttestlabresultsgt negative lt/
  testlabresultsgt
• lt/testlabanalysisResultsgt
• lt/soapbodygt
• lt/soapenvelopegt

19
Referring to Data ina P-Assertion

• ltsoapenvelopegt
• ltsoapheadergtlt/soapheadergt
• ltsoapbodygt
• lttestlabanalysisResultsgt
• lttestlabsampleIDgt Sample3
  lt/testlabsampleIDgt
• lttestlabresultsgt negative lt/
  testlabresultsgt
• lt/testlabanalysisResultsgt
• lt/soapbodygt
• lt/soapenvelopegt

20
Pointing to Data Within Assertions

• In order to refer to a part of an assertions
  content, we need to provide the following
• The ID of the assertion
• How to access the part within the assertions
  content

21
Pointing to Data Within Assertions

• In order to refer to a part of an assertions
  content, we need to provide the following
• The ID of the assertion
• How to access the part within the assertions
  content
• In the example, the ID of the assertion is given
  by
• Interaction Key
• View Kind
• Local P-Assertion ID

22
Pointing to Data Within Assertions

• In order to refer to a part of an assertions
  content, we need to provide the following
• The ID of the assertion
• How to access the part within the assertions
  content
• In the example, the ID of the assertion is given
  by
• Interaction Key
• View Kind
• Local P-Assertion ID
• The data part is given by a data accessor, such
  as an XPath expression
• / soapenvelope / soapbody / testlabanalysisResu
  lts / testlabsampleID

23
Related Interactions
Electronic Healthcare Record
Donor Data Collector
User Interface
Blood Analysis Results
Organ Donation Decision
24
Functional View of Decision
Lab Results
Healthcare Record
lttestlabresultsgt negative lt/ testlabresultsgt
ltechrsrecordgt lt/ echrsrecordgt
Diagnose Potential Organ
ltdonordecisiongt fit for donation
lt/donordecisiongt
Decision
25
Functional View of Decision
Lab Results
Healthcare Record
lttestlabresultsgt negative lt/ testlabresultsgt
ltechrsrecordgt lt/ echrsrecordgt
Diagnose Potential Organ
Parameter Names
ltdonordecisiongt fit for donation
lt/donordecisiongt
Decision
26
Functional View of Decision
Object
Object
Lab Results
Healthcare Record
lttestlabresultsgt negative lt/ testlabresultsgt
ltechrsrecordgt lt/ echrsrecordgt
Relation
Diagnose Potential Organ
Subject
ltdonordecisiongt fit for donation
lt/donordecisiongt
Decision
27
RelationshipP-Assertion
28
Sharing Context

• To connect assertions made by different actors
  some information must be shared.
• For example
• A sender receiver in an interaction need to use
  the same interaction key.
• An actor might like to inform other actors as to
  which provenance store it used to store
  assertions.
• Messages can be marked as being part of the same
  process.

29
Demarcating Processes
Run 1
Run 2
Testing Lab
Testing Lab
Donor Data Collector
Donor Data Collector
User Interface
User Interface
30
Demarcating Processes
Run 1
Run 2
Testing Lab
Testing Lab
Donor Data Collector
Donor Data Collector
Tracers
User Interface
User Interface
31
Tracers

• Tracers are added by an originating actor
• Tracers are passed between actors if the
  messages are part of the same process

32
Interaction Context

• States some metadata about an interaction.
• It contains
• An interaction key for that interaction
• And interaction metadata.
• Examples of such metadata are
• Tracers
• References to provenance stores that contain
  assertions about the interaction.

33
P-Header

• Interaction Contexts can be transferred in
  P-Headers.
• P-Headers are additional provenance specific
  information added to existing application
  messages.
• P-Headers can also be used to assign an
  interaction key to the message that its sent in.

34
P-Header
35
Message with P-Header

• ltsoapenvelopegt
• ltsoapheadergt
• ltpHeadergt
• ltinteractionKeygt
• ltmessageSourcegt Donor Data Collector
  lt/messageSourcegt
• ltmessageSinkgt Testing Lab
  lt/messageSinkgt
• ltinteractionIdgt 345 lt/interactionIdgt
• lt/interactionKeygt
• ltinteractionMetaDatagt
• ltlinkgt http//DDC.provenance.store
  lt/linkgt
• lttracergt http//process/9467 lt/tracergt
• lt/interactionMetaDatagt
• lt/pHeadergt
• lt/soapheadergt
• ltsoapbodygt lt/soapbodygt
• lt/soapenvelopegt

36
Summary

• Our data structures allow assertions about past
  processes to be succinctly and exactly specified
• Interaction and actor state p-assertions can be
  styled, so that what is asserted differs from
  that which actually occurred in a well-specified
  way
• Relationship p-assertions express functions
  between wholes or parts of p-assertions
• Actors share context to ensure that consistent
  identifiers are used throughout documentation
• Context can include process identifiers, called
  tracers

37
Questions?

• Simon Miles
• sm_at_ecs.soton.ac.uk