activity update
DESCRIPTION
Activity Update. WP4 Meeting Bologna – 29.07.2003 Simone Ludwig Electronic and Computer Engineering Department Brunel University / PPARC. Outline. Recent Work Basic Service Discovery Prototype Performance Measurements Ontology Design Rule-based Engine Planned/Ongoing Work - PowerPoint PPT PresentationTRANSCRIPT
Sem
an
tic S
erv
ice D
iscovery
Pro
toty
pe
Sem
an
tic S
erv
ice D
iscovery
Pro
toty
pe
Data
TA
GActivity UpdateActivity Update
WP4 Meeting Bologna – 29.07.2003
Simone LudwigSimone Ludwig
Electronic and Computer Engineering DepartmentElectronic and Computer Engineering Department
Brunel University / PPARCBrunel University / PPARC
DataTAG WP 4 Meeting, Bologna 2
Outline
Recent Work– Basic Service Discovery Prototype
– Performance Measurements
– Ontology Design
– Rule-based Engine
Planned/Ongoing Work– Integration of the semantic part with the basic service discovery prototype
– Resource Ontology
– Investigation of Similarity Matching
Time Outline
DataTAG WP 4 Meeting, Bologna 3
Architecture of Semantic Service Discovery Prototype
Matchmaking Engine
Service Request
Input/Output Process Resources
User Inter-face
User Inter-face
Service Registry (UDDI)
Service Registry (UDDI)
Grid Service
Ontology
Grid Service
Ontology
Service Response
DAML+OIL
Parser
DAML+OIL
Parser
Inference Engine (JESS)
Inference Engine (JESS)
Semantic Selection
Semantic Selection
Set of rules
Set of rules
Resource Ontology
Resource Ontology
Registry Selection
Registry Selection
Context Selection
Context Selection
HEP Applic.
Onotology
HEP Applic.
Onotology
DataTAG WP 4 Meeting, Bologna 4
Basic Service Discovery Prototype
Implementation of the basic service discovery prototype– OGSA-based
• XML• SOAP• WSDL• UDDI
GUI:http://193.62.142.4:31000/webapp/ServiceDiscoveryJSP/ServiceDiscovery.jsp
DataTAG WP 4 Meeting, Bologna 5
DataTAG WP 4 Meeting, Bologna 6
DataTAG WP 4 Meeting, Bologna 7
DataTAG WP 4 Meeting, Bologna 8
DataTAG WP 4 Meeting, Bologna 9
Performance Measurement Setup
3 different approaches– Centralised
– Decentralised
– Hybrid
DataTAG WP 4 Meeting, Bologna 10
Centralised Approach
Global Registry
VO1VO1 VO2VO2 VO3VO3
DataTAG WP 4 Meeting, Bologna 11
Measurements for CSD
136
137
138
139
140
141
142
143
WS1 WS2 WS3 WS4 WS5 WS6 WS7 WS8 WS9
SD
T /
ms
DataTAG WP 4 Meeting, Bologna 12
Decentralised Approach
Local Registry
Local Registry
Local Registry
RSDB RSDB RSDB
Or chain model
VO1VO1 VO2VO2 VO3VO3
DataTAG WP 4 Meeting, Bologna 13
Measurements for DSD
0
100
200
300
400
500
600
700
WS1 WS2 WS3 WS4 WS5 WS6 WS7 WS8 WS9
SD
T /
ms
DataTAG WP 4 Meeting, Bologna 14
Hybrid Approach
Global Registry
Local Registry
Local Registry
Local Registry
VO2VO2VO1VO1 VO3VO3
DataTAG WP 4 Meeting, Bologna 15
Measurements for HSD
0
100
200
300
400
500
600
Registry 1 Registry 2 Registry 3 Registry 4
SD
T /
ms
Service not found Service found
DataTAG WP 4 Meeting, Bologna 16
Comparison
0
100
200
300
400
500
600
700
CSD DSD HSD
SD
T /
ms
min. max.
DataTAG WP 4 Meeting, Bologna 17
Results
CSD DSD HSD
Admini-stration
Easy More difficult More difficult
Manage-ment
Easy More complex More complex
Security Easy More complex More complex
Scalability Not good Good Good
Perform-ance / SDT
Limited Good Good
Reliability Lowest Medium Highest
DataTAG WP 4 Meeting, Bologna 18
Ontology Design
Ontology Tool: Protégé
Application: HEP application use cases
Extraction of use cases -> ontology
-> HEP application ontology
DataTAG WP 4 Meeting, Bologna 19
DataTAG WP 4 Meeting, Bologna 20
Rule-based Engine
Also called Inference Engine Is a generic control mechanism that applies knowledge
present in the knowledge base (ontology) to task-specific data to arrive at some conclusion.
2 different approaches:– Forward chaining (data-directed inference):
• JRules• JESS
– Backward chaining (goal-directed inference):• Mandarax
DataTAG WP 4 Meeting, Bologna 21
Semantic Matchmaking Module
DataTAG WP 4 Meeting, Bologna 22
Integration
Integration of semantic part with basic service discovery prototype
Prototype will consist of:– Basic Part:
• Web/Grid services• SOAP• WSDL• Service Registry (UDDI)
– Semantic Part:• Context ontologies for the 4 HEP applications (CMS, ATLAS, ALICE, LHCb)• Grid Application Ontology• DAML+OIL Parser• Set of rules• Inference Engine
DataTAG WP 4 Meeting, Bologna 23
Resource Ontology
Extract the concept– Basic Structure of Resources
• CE• SE• WN• RB• UI
– Attributes of each resource element
– Relationship between the resources
Define the resource ontology
DataTAG WP 4 Meeting, Bologna 24
Time Outline
May June July August September October December
Basic SD Prototype
Perfor-mance
Measure-ments
Ontology Design
Inte-gration of semant. Part with basic SDP
Resource Ontology
(RO) Similarity Matching
Inte-gration with RO
November
DataTAG WP 4 Meeting, Bologna 25