Generic%20Interoperability%20Framework - PowerPoint PPT Presentation

About This Presentation
Title:

Generic%20Interoperability%20Framework

Description:

Long-term goal of information integration: independence of protocols, languages, data models and formats ... rdf:RDF xmlns:rdf='http://www.w3.org/TR/WD-rdf-syntax ... – PowerPoint PPT presentation

Number of Views:38
Avg rating:3.0/5.0
Slides: 30
Provided by: mel4
Category:

less

Transcript and Presenter's Notes

Title: Generic%20Interoperability%20Framework


1
Generic Interoperability Framework
  • Introducing a semantic-oriented middleware
    technology
  • Sergey Melnik, Stanford University

2
Motivation
  • Long-term goal of information integration
  • independence of protocols, languages, data models
    and formats
  • dynamic discovery of component functionality
  • universal interface

3
Universal interface
  • Evolvability
  • free well-defined extension
  • mixing, partial understanding
  • Interface description
  • Open, simple, efficient

4
Basic framework
  • Components exchange messages
  • Every message
  • uses generic representation
  • has well-defined meaning
  • Descriptions of all message parts are dynamically
    available

5
Generic representation
  • 1) Data
  • 2) Languages
  • 3) Protocols
  • 4) Interface descriptions, schemata etc.
  • directed labeled graphs (RDF)

6
Example protocols / message
7
Example serialization in XML
  • ltrdfRDF xmlnsrdfhttp//www.w3.org/TR/WD-rdf-sy
    ntax
  • xmlnsDLhttp//www-diglib.stanford.edu//Sear
    ch/Coregt
  • ltDLSearch queryDarwin AND (contraption OR
    machine)/gt
  • lt/rdfRDFgt

8
Example languages
9
Example data / metadata
10
Example schema / ontology
11
Evolvability Interoperability
12
Message composition
13
Message composition example
14
Layered architecture
15
Transport
  • Primitive message exchange only (message contains
    application-specific protocol information!)
  • Full-duplex communication channel
  • Serialization in XML (part of RDF standard)
  • Specifications for TCP/IP and HTTP

16
Layers API
  • interface Layer
  • Model dispatch(in Model m, in Layer from)
  • void connect(in String channel, in Layer l)
  • Generic dispatch
  • Model provides graph manipulation API

17
Example layers (1)
18
Connecting layers (1)
  • Layer l new LibraryLayer()
  • Layer t new TransportLayer()
  • l.connect( transport, t )
  • t.connect( dispatch, l )

19
Example layers (2)
20
Connecting layers (2)
  • Layer l new LibraryLayer()
  • Layer t new TransportLayer()
  • Layer e new ECommerceLayer()
  • l.connect( transport, e )
  • t.connect( dispatch, e )
  • e.connect( service, l )
  • e.connect( transport, t )

21
Unknown message
22
Connecting layers (3)
23
Connecting layers summary
  • no changes in Library components
  • universal E-commerce Layer
  • Can we do the same using distributed objects
    (CORBA)?
  • Dynamic Invocation and Skeleton Interface
    (DII/DSI) only!
  • Limitations interface descriptions, evolvability

24
Integration
  • Canonical wrappers
  • preserve semantics of sources
  • no additional functions
  • no common models or languages
  • Mediators
  • perform translation of protocols, languages and
    data

(simple!)
(? stateful graph transformation!)
25
Integration example (1)
  • Source 1
  • wind sensors
  • delivers list of vectors
  • Vector (x1,y1,x2,y2)
  • Source 2
  • interpolation based on energy conservation
  • computes map
  • 1 hour time to answer queries
  • Vector (x,y,?,l)
  • Mediator
  • real-time wind conditions
  • http//sfports.wr.usgs.gov/wind/
  • every hour
  • fetches sensor info
  • translates vectors
  • asks Source 2
  • forwards requests
  • Vector (x,y,?x,?y)

26
Integration example (2)
27
Current status
  • Java libraries
  • RDF parsing/serialization, model manipulation
  • schema evaluation subClassOf, subPropertyOf,
    validation
  • multithreaded TCP/IP, HTTP delivery
  • small footprint lt 90K !
  • Schemas (ontologies)
  • Core Communication, Search, Load Balancing, State
    Maintenance
  • http//www-diglib.stanford.edu/diglib/ginf/

28
Conclusion
  • Middleware for application development and
    integration
  • Ontologies of protocols, languages and data
  • Supports all of the semantics of components (not
    only querying!)

29
Future work
  • Modeling of advanced services
  • rights management, payment etc.
  • Sophisticated schema approaches
  • declarative, Turing-complete, etc.
  • On-demand layers, automatic mediation
Write a Comment
User Comments (0)
About PowerShow.com