Stanisaw Ambroszkiew icz the leader of the enTish team IPI PAN, - - PowerPoint PPT Presentation

stanis aw ambroszkiew icz
SMART_READER_LITE
LIVE PREVIEW

Stanisaw Ambroszkiew icz the leader of the enTish team IPI PAN, - - PowerPoint PPT Presentation

Stanisaw Ambroszkiew icz the leader of the enTish team IPI PAN, Polish Academy of Sciences and Institute of Informatics, University of Podlasie, Poland 1 According to SOA paradigm the clients requests y (tasks) are to be realized


slide-1
SLIDE 1

1

Stanis•aw Ambroszkiew icz

the leader of the enTish team IPI PAN, Polish Academy of Sciences and Institute of Informatics, University of Podlasie, Poland

slide-2
SLIDE 2

2

y According to SOA paradigm the clients’ requests

(tasks) are to be realized whereas the services are to be used

y What is service composition for? y

For creating sophisticaticated business procesess (BPEL4WS, WSCI, BPML, DAML-S, SELF-SERV) – procedural approach (How to do?)

y y

service architecture corresponds service architecture corresponds to SOAP + WSDL to SOAP + WSDL

y

For realizing clients’ task (SWORD, XSRL, enTish)- declarative approach (What to do?)

y

the task is specified explicitely; a protocol for task realization is needed

y y

more sophisticated service architecture is required more sophisticated service architecture is required

slide-3
SLIDE 3

3

for declarative approach

z The second layer has two interrelated components: raw

application (Function), and Filter.

z Raw application implements a single input-output operation. z Given a specification of the desired output to be produced by

the raw application (Function), the Filter replies with specifications of the required input.

slide-4
SLIDE 4

4

c clients lients’ ’ tasks tasks invocation, coordination, transaction discovery publication

Service requestor Service provider

serviceRegistry serviceRegistry Services Services ( ( applications applications) )

The challenge: How to realize such middlew are?

m middleware iddleware (broker) (broker)

Send task and get result register application

slide-5
SLIDE 5

5

middlew are

You can create a task, deliver input resources, and get the final result You can define your own data types, functions, relations; i.e., you own

  • ntology

You can join your application as service via API

slide-6
SLIDE 6

6

y the idea:

y service is represented by service-agent y client’s task is represented by task-agent y the agents speak one common language according to a conversation protocol y the conversation protocol is for service publication, discovery, composition, invocation, and transactions y

the basic components of the middleware:

y agents and y conversation language

slide-7
SLIDE 7

7

common State schema for task-agent and service-agent:

y Goal, I ntentions, Commitments, Knowledge

Message: z Header:

y From, To, Protocol, Order

z Body:

y a list of evaluated formulas of the contents language

state.xsd, message.xsd

slide-8
SLIDE 8

8

formula.xsd

z Contents language is a simple version of typed FOL without quantifiers:

y all names in the language are URIs that point to

concrete data

y names for types, functions, relations, variables y terms and formulas are defined in the standard way in the schema formula.xsd

z evaluated formula is defined in the schema info.xsd:

y formula y time&place stamp y signature (optional in the current version)

slide-9
SLIDE 9

9

definitions.xsd

z contents language is open and eXtensible:

y you can introduce your own ontology as an instance of definitions.xsd, i.e., introduce new types, new functions, and new relations to the language z upper ontology for sdc:

y properEntish.xml is an instance of definitions.xsd

y basic primitive concepts: agent, service, resource, intentions, commitments, timeout, ... z formula examples: y task formula: ?z= f(?x, g(?y)) and timeout(t0) y intention formula: • implies intentions( agent0 )

slide-10
SLIDE 10

10

y agents and services exchange messages with specific contents in

  • rder to realize:

y

service publication

y

service discovery

y

arranging services into a workflow

y

worklow execution and control

y

distributed transaction

slide-11
SLIDE 11

11

z workflow composition phase – the idea: y agent sends message to service0: „my intention is • ” y service replies: „I commit to realize • if you realize • ” y „• ” becomes the next intention of the agent y agent is looking for a service that can realize „• ” y suppose service1 could realize „• ” y agent sends message to the service1: „my intention is • ” y and so on ... more in enTish-Docs.pdf

slide-12
SLIDE 12

12

y the first version of enTish specs was completed (see enTish-DOCUMENTATI ON.pdf) y enTish is a working technology ! y prototype was realized (October 2002), is ready for testing and evaluation via three www interfaces starting with www.ipipan.waw.pl/ mas/ y work in progress on the next independent implementations

say nothing that isn’t worth saying