virtual enterprise synthesys

19
VIRTUAL ENTERPRISE SYNTHESIS BY WEB SERVICES COMPOSITION WITH HTN-LIKE PARSING ALGORITHM Victor Romanov, Ekaterina Yudakova, Svetlana Efimova, Alexandra Varfolomeeva 2015 SUMMER SIMULATION MULTI-CONFERENCE (SUMMERSIM’15) JULY 26-29, 2015 PALMER HOUSE HILTON CHICAGO, IL, USA

Upload: victor-romanov

Post on 12-Apr-2017

313 views

Category:

Economy & Finance


0 download

TRANSCRIPT

VIRTUAL ENTERPRISE SYNTHESIS BY WEB SERVICES COMPOSITION WITH HTN-LIKE PARSING ALGORITHM

Victor Romanov, Ekaterina Yudakova, Svetlana Efimova, Alexandra Varfolomeeva

2015 SUMMER SIMULATION MULTI-CONFERENCE (SUMMERSIM’15)JULY 26-29, 2015

PALMER HOUSE HILTONCHICAGO, IL, USA

Virtual Enterprise as a Global Trend

• Almost everything we use today is produced by a virtual enterprise composed of a number of companies.

• A virtual enterprise is a temporary alliance of cooperating enterprises that acts as a single organization to share skills, core competencies and resources in order to react quickly to changes in the market conditions and become agile enterprises. This collaboration is carried out using computer networks and web services.

• Collaboration of small enterprises gives them the chance to provide effective and individual customer service, which they couldn’t provide separately. Moreover, they get access to the bigger capital and various technologies due to cooperation with large organizations. Large enterprises serve customers faster and become more flexible.

Virtual Enterprise Structure

Activity_M

Business Partner_1

Business Partner_2

End User

Static Connection

Activity Type

Business Process

Service Request

Service Response

Dynamic Connection

Service Catalogue (UDDI)

Web ServicesWeb Service is, therefore, any service that:• Is available over the Internet or private (intranet)

networks• Uses a standardized XML messaging system• Is not tied to any one operating system or programming

language• Is self-describing via a common XML grammar• Is discoverable via a simple find mechanismA web service enables communication among various applications by using open standards such as HTML, XML, WSDL, and SOAP.

Web Services Provision Model

Universal Service Registries-Public UDDU-Private UDDI

Web Service Providers

Web Service Consumers

Role Operation Data Flows Message

Call

Link

WSDL link

WSDL fileSOAP request

SOAP response

Main Stages of Virtual Enterprise Development

• Stage 1: Elaborating requirements to the system being developed and its architecture.

• Stage 2: Translating the requirements into the formal language. Ontologies are used to declare and describe services.

• Stage 3: Designing the standardized description for searching web services based on SOAP, WSDL, OWL-S.

• Stages 4-5: Publishing/finding web service descriptions in the global UDDI registry. UDDI is an XML-based standard for describing, publishing, and finding web services.

• Stage 6: Virtual enterprise composition of obtained services.

Composition and Automated Planning• The most of composition approaches are

based on the automated planning algorithms.• The problem of automated planning is not

new, research in this domain are continuing for more than 50 years (STRIPS, HTN algorithms, automated planning languages ADL, PDDL).

• The languages of planning are based on the situation calculus, that is some extension of the first order predicate calculus.

The Statement of the Planning Task

• Let L - be first order predicate language, f – sentence (formula) at L. We will denote the set of all sentences by F. Let some subset of sentences is describing the definite state of some object. So the state s is just the set of sentences

• Let we have initial state and goal state Then we have some set of actions A. The action a A has a name a(ξ), precondition pre(a), positive post condition add(a), and negative post condition del(a).

• Post condition defines the result of action execution and situation change. Positive post condition adds new true sentence to the situation description and the negative one deletes true sentence from the situation description.

• Therefore action is a tuple of the next elements:{a(ξ), pre(a), add(a), del(a)}.

• Let y be transition function from state s to state s’ as the result of action a: s’=y(s,a).

• Then formal definition of the planning task is: find a sequence of actions and states, such as ), ),…) and

Values Assignment to the Parameters of the Actions

• Web service description , obtained by user from UDDI registry, is translated into a definition of the operator , on the base of an OWL-S Profile of the action

name (Ai ) = WSDiID.

}}{}{11

n

kki

n

kki itionhasPrecondWSDhasInputWSD

pre(Ai)

}}{}{ 11

n

kki

n

kki hasEffectWSDhasOutputWSD

add(Ai)

}}{1

n

kki hasEffectWSD

del(Ai)

STRIPS and HTN AlgorithmsSTRIPS(A, s, g) *p = empty planloop…if s satisfies g then return pa = [an applicable action in A, relevant for g]if a = null, then return failurep’ = STRIPS(A, s, precond(a))if p’ = failure, then return failures = apply p’ to ss = apply a to sp = p + p’ + a

The HTN-algorithm differs from classical in that here besides from elementary operators the group operators are created, named as tasks and by such way hierarchical network of tasks is formed. This approach reminds us the theory of formal grammars with its terminal and non terminal string and possibility of decreasing search process by excluding the combination of symbols belonging to the different categories.

*http://www.sharprobotica.com/2011/04/strips-for-classical-plan-representation-and-planning/STRIPS for Classical Plan Representation and Planning

Representation of the Virtual Enterprise Structure as a Parsing Tree

S

&

TR ST

&

TH FLTA

&

AC FACT

Å

lt lpt

Å

dt dpt

&

pc leb

sh SPex ci

su wsd

r bih

Terminal and Non-terminal Symbols of the Travel Agency Formal GrammarThe atomic services (terminal symbols) are introduced by lowercase letters:

• “lt” is for local taxi atomic service;

• “lpt” is for local public transport atomic service;

• “dt” is for destination taxi atomic service;

• “dpt” is for destination public transport atomic service;

• “pc” is for passport control atomic service;

• “sh” is for shopping atomic service;

• “ex” is for excursions atomic service;

• “d” is for diving atomic service;

• “ws” is for water sports atomic service;

• “ci” is for check-in atomic service;

• “r” is for restaurant atomic service;

• “bih” is for breakfast in the hotel atomic service.

The composite services (nonterminal symbols) are introduced by uppercase letters:• “S” is for tour package composite services;

• “TR” is for transportation composite services;

• “TA” is for transportation to the airport composite services;

• “TH” is for transportation to the hotel composite services;

• “FL” is for flight composite services;

• “ST” is for stay composite services;

• “F” is for food composite services;

• “AC” is for accommodation composite services;

• “ACT” is for activity composite services;

• “SP” sport composite services;

Formal Grammar Rules for Travel Agency

Formal grammar for travel agency can be described in the following way:•S → TR & ST•TR → TA & TH & FL•TA → lt Å lpt•TH → dt Å dpt•FL → pc & b & le•ST → ACT & AC & F•F → r bih•AC → ci•ACT → sh ex SP•SP → su d ws

The connection signs are used in the following meanings:

→ (imply) substitution (inference) rule& – (and) all symbols must enter to concatenation.Å – (xor) any one symbol may enter to concatenation. – (or) one or more symbols may enter to concatenation.

HTN-like Algorithm of Web Services Composition

1. The string of terminal symbols appears on the input stack. The string begins and ends with the sign“#”.

2. The stack is connected with recognizers of terminal symbols and recognizers of nonterminal symbols.

3. The recognizers of terminal symbols associate with terminal symbol (or sequence of terminal symbols) with one of the finite number of the categories of nonterminal symbols.

4. Every recognizer of terminal symbols has memory in the form of the list of terminal symbols that belong to the certain category. Every recognizer compares a terminal symbol in the stack with every terminal symbol stored in its list.

5. One or several symbols are read and recognized by the recognizer according to grammar rules as right part and replaced by left hand part in the stack for subsequent analysis and recognition.

6. According to the type of recognizer (&, Å, ) it looks for all, only one or several low level components.

7. The algorithm ends when the upper-level symbol S is reached.

TR is for transportation atomic and composite services

“TA” is for transportation to the airport composite services;“TH” is for transportation to the hotel composite services;“FL” is for flight composite services;“It” is for local taxi atomic service;“lpt” is for local public transport atomic service;“dt” is for destination taxi atomic service;“dpt” is for destination public transport atomic service;“pc” is for passport control atomic service;“b” is for boarding atomic service;“le” is for luggage examination atomic service.

Left Part of the Parsing Tree of Virtual Enterprise

TR

&

TH FLTA

Å

lt lpt

Å

dt dpt

&

pc leb

How The Bottom-up Recognizer Accepts The Input String While Parsing

(#ˇit lpt dpt pc b le#)(#itˇlpt dpt pc b le#) CURSOR SHIFT(#TA lptˇdpt pc b le#) XOR REDUCE, 2(#TAˇdpt pc b le#) XOR ABSORPTION,2(#TA dptˇpc b le#) CURSOR SHIFT,3(#TA THˇpc b le#) XOR REDUCE, 3(#TA TH pcˇb le#) CURSOR SHIFT,4 (#TA TH pc b le#) CURSOR SHIFT,4(#TA TH pc b leˇ#) CURSOR SHIFT,4(#TA TH FLˇ#) AND REDUCE,4(#TR#) AND REDUCE,1STRING IS ACCEPTED, SUBSYSTEM OF VE IS CREATED

THE INPUT STRING1. TR → TA & TH & FL2. TA → lt Å lpt3. TH → dt Å dpt4. FL → pc & b & le

TR

&

TH FLTA

Å

lt lpt

Å

dt dpt

&

pc leb

Conclusion• The rapidly changing business environment dictates the need for enterprises to

change their structures, processes and systems with the same speed in order to save clients and to remain competitive and compliant to the world around. The virtual enterprises provide such opportunity.

• The limited functionality offered by an atomic web service or stable set of services provided by traditional EIS (Enterprise Information System) providers cannot usually satisfy complex and rapidly changing customer requirements and appropriately reflect complicated business processes of virtual enterprises.

• As a great number of web services are available through the Internet, EIS providers should consider the new dynamic approach for building information systems based on web services composition in order to avoid the loss of significant part of potential customers and to satisfy the demand of different categories of users.

• Automated planning algorithms are important part of such synthesis.• In this paper, we have suggested to apply the theory of the formal grammar and

the appropriate procedures as means of automated planning support for web services composition.

Q&A

Thank you for your attention!

VIRTUAL ENTERPRISE SYNTHESIS BY WEB SERVICES COMPOSITION WITH HTN-LIKE PARSING ALGORITHM

Victor Romanov, Ekaterina Yudakova, Svetlana Efimova, Alexandra Varfolomeeva

2015 SUMMER SIMULATION MULTI-CONFERENCE (SUMMERSIM’15)JULY 26-29, 2015

PALMER HOUSE HILTONCHICAGO, IL, USA