Upload
santa
View
41
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Semantic Execution Meets Geospatial Web Services: A Pilot Application. Raluca Zaharia, Laurentiu Vasiliu (DERI) Joerg Hoffman (SAP Research) Eva Klien (Fraunhofer Institute). Terra Cognita 2008, Oct. 26, Karlsruhe, Germany. Overview. Motivation Scenario Semantic model Execution - PowerPoint PPT Presentation
Citation preview
Copyright 2008 Digital Enterprise Research Institute. All rights reserved.
Digital Enterprise Research Institute www.deri.ie
Semantic Execution Meets Geospatial Web Services:
A Pilot Application
Raluca Zaharia, Laurentiu Vasiliu (DERI)Joerg Hoffman (SAP Research) Eva Klien (Fraunhofer Institute)
Terra Cognita 2008,
Oct. 26, Karlsruhe,
Germany
2 of XYZ
Digital Enterprise Research Institute www.deri.ie
Overview
Motivation Scenario Semantic model Execution Results Conclusions
3 of XYZ
Digital Enterprise Research Institute www.deri.ie
Motivation
Why semantics and geospatial web services:
data sharing and processing functionalities for geospatial entities
not just one service metadata standards and catalogue services for
description and search of geospatial services no specifications to formally and explicitly define
semantics of the data and functionality
4 of XYZ
Digital Enterprise Research Institute www.deri.ie
Motivation
We want automation Discovery Interoperability Integration
A semantic framework (WSMO/L/X) semantic descriptions modelled unsatisfactory execution performance
complex scenario large amount of data
of geospatial services
5 of XYZ
Digital Enterprise Research Institute www.deri.ie
Scenario
DepartmentsDepartments
ConsumptionConsumption
Production-Consumption
Map data
Production-Consumption
Map data
QuarriesQuarries
PopulationPopulation ConsumptionConsumption
ProductionProduction
Average consumption
Average consumption
PopulationPopulation
ProductionProductionQuarriesQuarries(lat, long)
(lat, long)
6 of XYZ
Digital Enterprise Research Institute www.deri.ie
Mediators
Semantic Model
OntologiesOntologies
GoalsGoals Web servicesWeb services
7 of XYZ
Digital Enterprise Research Institute www.deri.ie
Modeling tools
WSMT v1.4.1 CompositionS
tudio
8 of XYZ
Digital Enterprise Research Institute www.deri.ie
Execution
Co
mm
un
ica
tio
n
Ma
na
ge
rC
om
mu
nic
ati
on
M
an
ag
er
Pa
rse
rP
ars
er
Dis
co
ve
ryD
isc
ov
ery
Da
ta M
ed
iati
on
Da
ta M
ed
iati
on
Ch
ore
og
rap
hy
E
ng
ine
Ch
ore
og
rap
hy
E
ng
ine
Inv
ok
er
Inv
ok
er
CoreCore
Resource Manager
Resource Manager
WSMO2Reasoner Framework
WSMO2Reasoner Framework
WSMX
AdminWFSAdminWFS
QuarriesWFSQuarriesWFS
PassthroughPassthrough
Geospatial Web
Services
9 of XYZ
Digital Enterprise Research Institute www.deri.ie
Execution
WSMX Execution Semantics
WSMX Choreography
Engine
Requester Choreography
Provider Choreography
WSMX Invoker component
registerChoreography (goal) initialize (state signature, choreography rules)
registerChoreography (WS) initialize (state signature, choreography rules)
updateState (R-to-P, instances) update (instances)
step
instances to send
instances to send > 0
invokeWS (instance to send)
service response ontology
create response ontology
updateState (P-to-R, instances) update (instances)
step
isProviderChorInEndState()
in Choreography Execution
10 of XYZ
Digital Enterprise Research Institute www.deri.ie
Improving the performance
Generality vs. performance
Minimum number of steps Minimum number of instances in the state
ontology -> call services early -> eliminate redundant calls -> cache reasoning results
11 of XYZ
Digital Enterprise Research Institute www.deri.ie
Results
Surface sizeWeb services
(Total duration)
Instancesin the state
ontology
Currentimplementation
with reasoning tool
KAON2 IRIS
1 department 7 WS (3.84s) 164 88.57s 25.43s
2 departments 12 WS (5.03s) 321 184.5s 41.48s
3 departments 17 WS (6.89s) 430 314.37s 55.78s
15 department
s
77 WS (33.89s) 2082 4531.20s 492.41s
12 of XYZ
Digital Enterprise Research Institute www.deri.ie
Conclusions
Real-life scenario of semantic execution in the geospatial domain
Why use semantics (the framework supports): Semantically described elements are a lot easier
to (re)use Reduced implementation time Increased agility N:M integration
13 of XYZ
Digital Enterprise Research Institute www.deri.ie
Conclusions
Degree of automation Services ranked and selected accurately Compositions more adaptable to changes
User input/approval still required.
Automatic composition not available(lack of semantic descriptions, tools for users etc.)
14 of XYZ
Digital Enterprise Research Institute www.deri.ie
Conclusions
Support for creation and execution of complex and flexible compositions
Service requester and service provider completely decoupled
An approach for generating the composition rules can used on top of the existing framework
ASM: simple and intuitive way of defining service compositions
“describe the requirements rather than telling the system step by step what to do”
15 of XYZ
Digital Enterprise Research Institute www.deri.ie
Conclusions
Each domain has particular requirements => discovery, mediation, execution etc. stressed in
a different way
SWS techniques over real services from the geospatial domain => huge execution overhead introduced
Significant advantages of the approach => mandatory to improve the framework performance.