Grid Service Discovery with Grid Service Discovery with Rough Sets Rough Sets Maozhen Li, Member, IEEE, Bin Yu, Omer Ran Maozhen Li, Member, IEEE, Bin Yu, Omer Ran a, and Zidong Wang, Senior Member, IEEE a, and Zidong Wang, Senior Member, IEEE IEEE TRANSACTION S ON KNOLEDGE AND DATA EN IEEE TRANSACTION S ON KNOLEDGE AND DATA EN GINEERING, VOL. 20, NO. 6, JUNE 2008 GINEERING, VOL. 20, NO. 6, JUNE 2008 Present by Chen, Present by Chen, Ting-Wei Ting-Wei
32
Embed
Grid Service Discovery with Rough Sets Maozhen Li, Member, IEEE, Bin Yu, Omer Rana, and Zidong Wang, Senior Member, IEEE IEEE TRANSACTION S ON KNOLEDGE.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Grid Service Discovery Grid Service Discovery with Rough Setswith Rough Sets
Maozhen Li, Member, IEEE, Bin Yu, Omer Rana, and ZidMaozhen Li, Member, IEEE, Bin Yu, Omer Rana, and Zidong Wang, Senior Member, IEEEong Wang, Senior Member, IEEE
IEEE TRANSACTION S ON KNOLEDGE AND DATA ENIEEE TRANSACTION S ON KNOLEDGE AND DATA ENGINEERING, VOL. 20, NO. 6, JUNE 2008GINEERING, VOL. 20, NO. 6, JUNE 2008
Present by Chen, Ting-WeiPresent by Chen, Ting-Wei
2
Outline
Introduction The Design of ROSSE QoS Modeling ROSSE Case Study ROSSE Evaluation Conclusions
3
Introduction
ROSSE • Rough sets-based search engine
• Discovery Grid service
• Maximize user satisfaction in service discovery
Evaluate the discovery of computing services • Accuracy
• Efficiency
4
The Design of ROSSE (cont.)
Service publication Service discovery
5
The Design of ROSSE (cont.)
6
The Design of ROSSE (cont.)
Step 1• advertise the service to ROSSE through a
Web user interface
Step 2• Load into the ROSSE Service Repository
• Names
• Properties
7
The Design of ROSSE (cont.)
Step 3• Publish service ontology that can be defined
in OWL
Step 4• Load into the ROSSE Ontology Repository
• Inference engine to infer the semantic relationships of properties
8
The Design of ROSSE (cont.)
Step 5-6
Step 7-9
Step 10-11
Step 12-13
Step 14-16
9
The Design of ROSSE (cont.)
Step 5• Post a service query to ROSSE
• Service category of interest
• Expected service properties
• Via its Web user interface
Step 6• Pass to the Irrelevant Property Identification c
omponent
To page 8
10
The Design of ROSSE (cont.)
Step 7• Access the ROSSE Service Repository
Step 8 • Identify and mark the properties of advertised
services
• Define in the ROSSE Ontology Repository
Step 9• The query is passed to the DPR component
To page 8
11
The Design of ROSSE (cont.)
Step 10• Access the ROSSE Service Repository to
identify and mark dependent properties
Step 11• The DPR component invokes the Service
Similarity Computing (SSC)
To page 8
12
The Design of ROSSE (cont.)
Step 12• Access ROSSE Service Repository
• Compute the match degrees of relevant properties of advertised service to the service query
Step 13 • Functionally matched services have distinct
nonfunctional properties related to QoS
• SSC invoke the QoS ModelingTo page 8
13
The Design of ROSSE (cont.)
Step 14• In turn filters functionally matched services
Step 15• Via the Web user interface of ROSSE
Step 16• A list of discovered services
To page 8
14
The Design of ROSSE (cont.)
Rough Sets for Service Discovery• Mathematical technique to deal with uncertainty in
knowledge discovery
• Rough set theory
• definitely has property p
• possibly has property p
• absolutely does not have property p
{ :[ ] }
{ :[ ] 0}
A
A
P
P
X x U x X
X x U x X
,x X x
,x X x
,x U X x
15
The Design of ROSSE (cont.)
• Rough set theory for ROSSE
X
1X
2 2:X X X X
1 0X
1 0X
X 1X
2X
1 0X
1 0X
1X7
8
16
The Design of ROSSE (cont.)
Irrelevant Property Identification• Semantic relationships with the properties
• Define• Exact match: pQ=pA, or pQ is a subclass of pA
• Plug-in match: pA subsumes pQ
• Subsume match: pQ subsumes pA
• Nomatch: No subsumption between pQ and pA
• Uncertain: No subsumption between pQ and pA, and pA=NULL