Knowledge Creation Tools for DAML

1
Knowledge Creation Tools for DAML
Grit Denker, Jerry R. Hobbs, David MartinSrini
Narayanan, Richard Waldinger
SRI International
2
Outline

• DAML-S A Service Markup Language.
• Process Models for Web-services
• Core theories and markup for transactions,
  processes, time, interactions
• Execution Semantics, monitoring, and simulation--
  DEMO
• Other Work-in-progress
• Theories inference for the Semantic Web
• Models of Security and Trust
• Resource Site Construction Tool
• DAMLOIL plugin to Protégé DEMO
• FrameNet in DAML

3
DAML-S Goals

• Full automation of service use
• DAML markups provide enough info for an agent to
  find, select, enact and use a service never
  before encountered
• Service requests handled seamlessly with
  information queries
• Allow for composition of both
• Many components tools can work for both
• Search selection, ontology translation,
• Support inference in selecting and using services
• Joint work with Stanford-KSL, CMU, BBN, Nokia

4
Service Ontology Top-level Classes
Service
Resource
provides
presents
supports
DescribedBy
ServiceProfile
ServiceGrounding
ServiceModel
What theservice does
How to access it
How it works
5
Process Upper Ontology
6
Time Ontology
Intervals
during
startOf
endOf
Process
inside
atTime
Instants
before
Future Developments Temporal Measurement
Clock and Calendar
7
(Simple) ProcessWhat does it do?

• Requirements for use results of use
• Black box view Information needed to execute
  an atomic service
• Inputs, outputs, preconditions, effects,
• Binding rules for inputs, outputs
• Roles involved
• May vary for different service classes
• Can employ logical rules, conditional I/0,
  effects.
• Atomic Transaction

8
Service ModelHow does it work?

• Semantic description of a service
• Glass box view
• Detailed characterization of what it does
• May vary for different service classes
• Analogous to procedure body (but abstract)

9
B2C Purchase ProcessModel
Locate-Goods
Compositeservice
Put-in-Cart
Check-out
Sign-in
Create-Account
Load-Prefs
Create-Prefs
Simple or compositeservices
One-Step
Select-Payment-Method
Specify-Delivery-Address
Giftwrap
Finalize
10
Congo.daml (partial)
ltrdfsClass rdfID"CongoBuyBook"gt
ltdamlintersectionOf rdfparseType"damlcollectio
n"gt ltdamlClass rdfabout"processSequence
"/gt ltdamlRestrictiongt
ltdamlonProperty rdfresource"processcomponents"
/gt ltdamltoClassgt
ltprocess-seqOf- rdfparseType"damlcollection"gt
ltrdfsClass
rdfabout"TopLevelBuyChoice"/gt
ltrdfsClass rdfabout"SpecifyDeliveryDetails"
/gt ltrdfsClass
rdfabout"FinalizeBuy"/gt
lt/process-seqOf-gt lt/damltoClassgt
lt/damlRestrictiongt lt/damlintersectionOfgt lt/
rdfsClassgt
11
Process Model and Inference

• Model
• Graphical Model is constructed recursively from
  DAML Markup
• Model is stochastic and can simulate execution
• Integrated into OAA and SNARK
• Inference
• Verfication, deadlock, reachability analysis,
  and performance analysis (such as
  forward-backward computing Viterbi paths).

12
Sequence(T1,T2)
finish
start
Ongoing
P3
P1
T2
P2
T1
13
concurrent (T1, T2, T3)
P1
T1
P2
start
T2
finish
P3
T3
bypass
14
concur_sync (T1, T2, T3)
P1
T1
P1
P2
start
T2
P1
finish
P3
T3
P1
15
unordered (T1, T2, T3)
P1
T1
P2
start
T2
finish
P3
T3
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
Execution Semantics

• Model is a high level Stochastic Petri Net and
  has a well specified execution semantics.
• Being extended to the more expressive Rewriting
  Logic (joint with Jose Meseguer)
• Rewriting Logic is a logic of distributed
  systems.
• There is a widely used rewriting logic tool,
  Maude. More at http//maude.csl.sri.com

21
Status

• Process Tool ready for release , pending stable
  DAML-S.
• DAML-S 0.5 at http//www.daml.org/services/
• Joint work with Stanford KSL, CMU, BBN, Yale,
  Nokia
• SRIs DAML work is described at
  http//www.ai.sri.com/daml

22
Whats Hot

• DAML-S is a serious attempt to use DAMLOIL.
• Process Modeling Tool for DAML-S descriptions
• Execution Semantics
• Resources,
• Execution monitoring
• Simulation
• Inference

23
Whats hard

• DAML-S expressivity/naturalness issues
• How to define scripts
• How to define unification constraints, etc..
• Services are like verbs, ontologies are more
  natural to represent nouns.
• Achieving and maintaining consensus
• Coordinating with industry standards
• With existing industry proposals
• ex. XLANG