Semantic Web Services Composition via Planning as Model Checking

Hong Qing Yu and Dr. Stephan Reiff-MarganiecComputer Science Department

Introduction

Background and the framework of SWSC Case study Web Services model Planning as model checking Advantages and future work

Background of SWS

Syntax only!

Web Service Architecture

WS standards: Lack of semantics!

Semantic Web Services What should S+WS ontologies provide?

(Mainly) Automation of the Usage Process:

Publication

Discovery

Selection

Composition

Execution

Monitoring

The framework of SWSCPhase 1 : Specification

Specify the Planning Goal

Provide the initial situation

Phase 2 : Model extraction

Select WS which in the plan domain

Extract WS models

WS Repository

OntologiesPhase 3 : Planning

Phase 4 : Physical Composition & Execution

Selection

Generation

Execution

Case study of Web Services Composition

WS3= TV shop (S)

WS4= Item delivery (D)

WS2= TV Information

WS5= Insurance (I)

WS1=Locate IP

&& After

WS6= TV License

Sm

art P

ortal

Initial Situation

Goal

S (D I)

Services

WS model

Operation Name

Precondition

State

Input message (Parameters)

Output message (Parameters)

Communications

(Type, Role)

Domain

Purpose

Quality

Operation Operation Operation Operation

WS model

WS1

Located

Mout (string Location)

E-shopping

Locating

high

Min (string IP_address)

WS2

Got_infor

Mout (string Brand, string Type, double S_size,

string review, Colo_type)

E-shopping

TV_infor

high

WS3

Purchased

Mout (string S_adress, double value, double

TV_size)

E-shopping

TV_sell

high

Min (string Brand, double S_size, string Type, string

Location, string TV_license )

WS4

Deliveried_Item

E-shopping

Delivery_item

high

Min (string S_address string Location, double

size)

WS5

Bought_insurance

Mout (string reference)

E-shopping

Insurance

high

Min (double value, string C_address, string

Goods_type)

WS6

Got_TVL

Mout (string TV_license)

E-shopping

TV_license

high

Min (string C_address, string Colo_type)

Got_TVL

Purchased Located

Mout (date delivery_time, double cost)

SelectConfirm

Request

Confirm

Request Request

Select Confirm

Request

Composition problem model

Specification for the goal

Specification for start conditions and data 1. We are planning from initial operation state2. The initial knowledge is the information which submitted by Client user

Our case1. Initial state is start2. Initial knowledge is Customer address, Goods type (TV), IP address

S (D I)

Planning as Model Checking

sState: {Start,

Parameter: {string C_address, string Goods_type, string IP_address,

1 2

2

6

3

4 old

5 4

old

5

Located, Got_infor,

Got_TVL, Purchased, Delivered_item, Bought_insurance}

string Location,string Brand, string Type, double S_size, string review, Colo_type,string S_adress, double value,

double TV_size,date delivery_time, double cost,string reference}

string TV_license

1

old

S (D I)

Advantages and future work

Advantages: Not rely on any particular ontology language Simple specification Executable + Reusable

Future Work: More complex goals Add non-functional requirement to planning

algorithm Interleaving of services in plans Complete the framework

Thanks

Any Questions?

TR available : http://www.cs.le.ac.uk/people/hqy1/swsc_pamc1.0.pdf