manish(ECE)x - vsrd international journals division

stalksurveyorSecurity

Nov 3, 2013 (3 years and 7 months ago)

85 views

Context Aware Service Matchmaking Using Naïve Bayesian Classifier

*
Manish Saxena








Reserch Scholar








Electronics & Communication Engineering





CMJ University

Shillong


Email:
gocetinstitute01@gmail.com


**Dr.


S. K. Shrivastava

Principal

SBITM, Batul

1.

ABSTRACT

With enormous increase in the availability of Web services, matchmaking of relevant web services has become
a sign
ificant challenge in the process of service discovery. The current solutions available for service
matchmaking use keywords, through which only primitive matching is possible. These keyword based
mechanism provides irrelevant results and also do not have c
ontrol over all the available services. To cope with
this problem, this paper focus on classifying the available services and locate the relevant services based on the
context of information requested by customers. To automatically generate wrappers for th
e service description,
we used co
-
occurrence based Clustering algorithm and Naive Bayesian classifier model


.
Keywords :

Web services Matchmaking, Service selection, Service discovery, Naïve Bayes classifier,
Ontology.

2.

INTRODUCTION

Web services technology
allows applications to

communicate with each other in a platform and

programming
language independent manner for the

development of distributed computing based on SOC

(Service Oriented
Computing) and deliver application

functionalities as services to eithe
r end
-
user application

or other services.
These services are autonomous,

platform
-
independent entities that can be delivered,

published, discovered and
loosely coupled. With technologies such as SOAP [13] a simple XML
-
based

lightweight protocol to let
appl
ications exchange

information over HTTP, WSDL [14] an XML
-
based

language for describing Web
services and how to

access them. UDDI [15] is a repository for dynamic

service publication and discovery. Web
services are the

powerful mechanism for integrating ex
isting software

applications over the web, independently
of

programming language, execution platform or

transport protocol.

Web Services keep on emerging at an ever

increasing pace. As Web services increases, many

businesses are providing similar services
with

overlapping
functionalities. Therefore, it is very

difficult for users to find an appropriate service among

the sea of services.
Research in web services covers

various areas of web services starting from service

grounding to service
composition and s
ervice mining.

Web service discovery is one such area that deals with

the process of locating
or discovering related service

descriptions that describes a particular web service

using the Web Service
Description Language (WSDL).

Service matchmaking in web
service discovery is

concerned with matching the
service query

requirements with number of available services in the

registry.

To make the process of service
matchmaking in

implementing service discovery more appropriate,

service matchmaking can be automat
ed. Not
only the

automated system should produce a result, but it

should also do so in an efficient manner. Efficient here

means the context and functional quality of the

services as specified by the user as their preferences.

Context is
the information th
at characterizes the

interactions between humans, applications, and the

surrounding
environment. From a Web services

perspective, context is defined as a set of common

meta
-
data about the
current execution status of a Web

service and its capability of
collaborating with peers,

possibly enacted by
distinct providers or customers [1].

Here, context is considered as key information

characterizing the customer’s
information, e.g.,

customer preferences, needs of customer, about the

location, and useful infor
mation about the
environment

where customer operates, e.g., date, time, on
-
going

activities and interactions with services and/or
other

c
ustomers.

This paper proposes a matchmaking process to find

services by making use of context and semantics

embodied
in

the service query. To define and extract

services among available services in the registry, this

paper makes
use of domain ontology. Merely, to make

the process of service matchmaking easier, this paper

also focuses on
classifying the service description.

This

is achieved by automatically generating wrappers for

the service
description with the use of co
-
occurrence

based clustering algorithm and Naïve Bayesian

classification model.

The remainder of the paper is structured as follows:

Section 2 describes th
e related works on service

matchmaking process. The proposed matchmaking

process is described and illustrated in Section 3.

Section 4
covers the classification technique of service

description by naïve bayes classification model.

Finally, Section 5
gives t
he conclusion and future work


2.
LITERATURE SURVEY

Various techniques and approaches have been

proposed in the literature for service matchmaking to

implement
service discovery.

Weili Han, Xingdong and et al [2] proposed

context and semantics based matchm
aking where
the

semantics of both the technical and business process of

registered services were used to improve the
precision

of matchmaking in an integrated development

environment based on Eclipse. However, the context

information they used depends on t
he infrastructure

and was suitable for the closed system, so it is weak in

interoperability.

Sudhir Agarwal and Stephen Lamparter [3]

proposed a generic approach for modelling
aggregation

of information for various web services attributes as

part of
ontology for describing composite web

services. They have used a tool that allows a user to

enter his preferences and provide the user a list of

generated
services sorted by rank. Depending on these

services a user can decide more easily which of the

servi
ces, he
wishes to execute. Further, users can

define an earlier threshold of rank for the services,

which he wishes to be
executed. Kuster et.al proposed

[4] an approach to automate service discovery,

matchmaking and composition.
During the service

matchma
king, they use a single and multiple
-
effects to

rank the qualified services. Keild et al
[5] presented a

generic framework to support the development of

context
-
aware adaptable Web services. This

framework separates clients/Web services from the

context fr
amework that supports clients and services.

The
transfer of context information is performed

through SOAP message header. Context information

can be
explicitly and directly processed by clients or

Web services or be automatically handled by the

context
fra
mework.

The support for context
-
aware services depends on

an improved semantic model of services by
using

ontologies that support formal description and

reasoning [6]. Semantic model may contribute not only

to
handle problems related to service interoperab
ility,

but also in order to take into account different aspects

of the
environment in which the service is executed.

Yoji Yamato and Hiroshi Sunaga [7] proposed a

flexible service
composition framework, where a

semantic
-
level service situation is translate
d and its

components are
dynamically originate, elected, and

bound. Service components are changed with respect

to the context change
since user situations may change

instantly. Jian Wu and Zhaohui Wu [8] developed a

model by classifying the
properties of
web services

into four categories. These categories have been given

certain similarity assessment
methods and based on

these assessment methods; the matchmaking process

has been done on the web to
improve the effect of web

services. Here, WSDL specificatio
ns are developed

independently from the module
code.

Zakaria Maamar et al [9] proposed a technique

which takes into account both the context and policy

for
Web services composition. The role of context

provides information about environment the service

composition
occurs. The transition of one level to

another level produce the action taken place, the

context and type of policy
used. Three types of policies

are used in this technique namely Engagement,

Mediation and Deployment.
Policies are defined as

‘‘
information which can be used to modify the

behaviour of a system’’. The structure for
each type of

context is defined along with its arguments. The

composite level includes arguments such as label,

global ontology, previous component web service,

current
component, next component, beginning time,

and
date. The semantic kevel includes arguments such

as label, current composite web service, current component
web service, triggered mapping function per current composite web service, triggered

conversion funct
ion per
current component web

service, and date.

Most existing matchmaking techniques are intended

for tightly coupled systems, within the control of a

single
organization. The interaction model between

components in a Web service
-
based system is also

diff
erent from
that of components in a single

organization

environment. Therefore, it is clearly seen

from the literature that the
embracing of context
-
aware

with semantic matchmaking of services enhances the

performance to produce the
optimized solution when

compared with conventional service matchmaking

techniques


3.
SERVICE MATCHMAKING

Service matchmaking is the process of selecting

appropriate service(s) from a set of available services

for a
service request. In other words, when we check

whether a service

matches to a request, we actually

check
whether some functional and temporal properties

required by the consumer are satisfied by the service or

not. It
is executed only when the web service request

and web service advertisement matches. The main

issue in

matchmaking is how to represent the

advertised and requested services, and how to calculate

possibility
matching between these services. The

advertised and requested services can represent data or

services by using
many type of representation. Syntax

base
d matchmaking yields irrelevant retrieval of

services since nature of
service is not captured in the

service description. The traditional keyword
-
based

matchmaking supported by
UDDI registries only

considers inputs and outputs for matchmaking or

simplify t
he functional descriptions. For
example,

keywords extracted by many queries in the same

domain may not be considered with the same

importance, since some keywords may be more

significant than others. Therefore, the need to give a

different
interpretation a
ccording to the context

emerges. Here we propose a service matchmaking

model where the
context and functional properties of a

service are expressed in terms of inputs, preconditions

and effects. Input
terms are represented as concepts

with the use of domai
n ontologies. Domain ontology

provides a conceptual
structure of a problem domain

that holds information collected and processed in a

classified manner. Thus, it
eliminates the problem

caused by syntactic matchmaking.

In addition to these ontologies, conte
xt of the user

play a major role in matching the desired services.

The

pictorial representation of the

service matchmaking
process. A service retriever uses

the service lookup directory to access the services from

the UDDI repository
based on the service c
onsumer’s

request. It initially uses the semantic matchmaking to

select the appropriate
services. This can be done by

using ontologies, which would be discussed detailed in

later.

The retrieved
services are matched with respect to

contextual parameters suc
h as business and

environmental properties. The
business properties

includes such as cost, organization arrangement,

payment method and reputation. The
environmental

properties include the temporal and location of the

service. The service retriever selects

the
requested

services from the repository and uses the matchmaking

algorithm to select the related services with
respect to

the contextual parameters. Figure 1 clearly depicts the

service matchmaking process. The process
starts by

submitting the user
query to retrieve the services that

satisfy their needs and requirements. The Query

processor on the client side, process the query by

identifying the functional semantics and non functional

qualities embodied in the user query.

The service context extract
or analyzes the related

contexts of the service to be selected with aid of the

query
processor agent and prepares the contexts

information of the web service with the use of domain

ontology. On
the provider side, the service retriever

agent analyzes the se
rvices that are published by the

service provider and
evaluates the functionality of

services defined in the service specification. The

service specification holds
service id, service name,

providers name, provider’s id, service description, QoS

details et
c. The evaluated
services are then passed to

service context extractor. The extraction of service

context is a critical process which
is not focused in this

paper. The context extractor extracts the context of all

the services and is send to service
classi
fier. The

process of service description classification is

described in the next section.

Service Lookup directory on the client side is used

to select the services for matchmaking that has been

recently
used or retrieved by the user. It has all the

availa
ble services the user has retrieved recently.

Initially the service
retriever selects the services with

the intervention of service lookup. Here, the semantic

based retrieval is
supported by the ontology. Ontology

represents information about a certain dom
ain. This

information includes
entities in the domain, their

property and relationship with each other. Entities in

the ontology are termed as
Concepts. A well
-
defined

syntax is required to explicitly represent concepts of a

domain. RDF [10] context is
use
d for describing

ontology. Web Ontology Language (OWL) [11] is

developed on the top of RDF and is used
for ontology

description.

The service repository is where all the published

services are found and invoked across
the Web. Based

on XML, SOAP and other s
tandards for service

invocation allow an application to interact with
remote

applications to select the required services. In this

matchmaking process service repository is used to

access all the services requested by the consumer. The

service matchmaker
through the service lookup

directory
may consume the services and check for the

services with the requested and advertised service

description if it is
been already used. Otherwise, the

process of matchmaking is done by matching with the

classified service
s
description kept in the service

repository.

The services, which match exactly both the service

requester and service provider description, are

consumed by
the service consumer for further service

discovery and service composition process.



4.
CLASSIFICA
TIONS OF SERVICES

To make the process of service matchmaking easier,

this section describes the classification of service

description. Once the service description of the

published services is classified, the task of matching

the
semantics found in service

request to semantics of

available services in the repository can be easily done
.



4.1 FORMULATION OF PAIR
-
WISE CONSTRAINTS

Classification of service description reflects the

semantics of the services as well as its region and we

would like
to induce from

the concepts that

cannot_link
and
must_link
relations between different

regions. In general it is
easier to induce the
cannot_link

relationship from shared concepts but the
must_link

relationship is harder to
deduce. We assume that the

semantic
irrelevance of two regions can be deduced by

the irrelevance of all
concepts between two regions.

This assumption is reasonable because although

classification of description is
likely to be incomplete,

it is always complete for those concepts that we care

most.


4.2 CO


OCCURRENCE BASED CORRELATION

In general, high concurrence concepts are likely to

be used together to describe same service description.

In
other words, two services are likely to belong to the

same conceptual group if they have high co
-
occ
urrence

and
vice versa.



4.3 CLUSTERING USING PAIR
-
WISE CONSTRAINTS

After the pair
-
wise construction between two

regions of service description, we perform clustering to

generate
region clusters. K
-
Means is a popular

clustering method. Since K
-
Means cann
ot directly

handle pair

wise
constraints, we adapt a variant of

K
-
Means performing the clustering.


5. CONCLUSION

The matchmaking process should be efficient in service selection to locate and select the requested

services by
the consumer. This paper
uses both the

context and semantic property of services for

matchmaking. To make the
process of service

matchmaking easier, a service description classifier

technique is proposed.


REFERENCES


[1]

Z. Maamar, N.C. Narendrab, S. Sattanathan, “Towards an ontology
-

based approach for specifying and
securing Web services”, Journal of Information and Software Technology, Elsevier, vol: 48, pp. 441


455,
2006.

[2]

Weili Han, Xingdong Shi, Ronghua Chen, “Process
-
context aware matchmaking for web service composition”,
Inte
rnational Journal of Network and Computer Applications
, Vol
-
31 (2008), pp.559

576.

[3]

Sudhir Agarwal, Steen Lamparter “User Preference Based Automated Selection of Web Service Compositions”,
Workshop on Dynamic Web Processes
, pp. 1
-
12, 2005.

[4]

Issarny V, Caporu
scio M, Georgantas N, “A Perspective on the Future of Middleware
-
based Software
Engineering”,
In Proceedings of International Conference on Software Engineering
, 2007.

[5]

Kuster U, Konig
-
Ries B, Stern M, Klein M. Diane, “An integrated approach to automated service discovery,
matchmaking and composition”,
In Proceedings on International World Wide Web Conference
, Banff, Alberta,
Canada, pp. 1033

42, 2007.

[6]

Yoji Yamato, Hirosh
i Sunaga, “Context
-
Aware Service Composition and Component Change
-
over using
Semantic Web Techniques”,
In Proceedings of IEEE International conference on web services
, 2007.

[7]

Li Kuang, Jian Wu, Shuiguang Deng, Ying Li, Wei Shi, Zhaohui Wu, “Exploring Semantic Technologies in
Service Matchmaking”,
In Proceedings of IEEE International conference on web services (ICWS 2005).

[8]

Zakaria Maamar, Djamal Benslimane, Philippe Thiran, Chir
ine Ghedira, Schahram Dustdar, Subramanian
Sattanathan, “Towards a contextbased multi
-
type policy approach for Web services composition”,
Journal on
Data and Knowledge Engineering
, Elsevier, Vol. 62, pp: 327


351, 2007.

[9]

Francesco M, Donini Tommaso Di Noia
, Eugenio Di Sciascio, Marina Mongiello, “Semantic matchmaking in a
p2p electronic marketplace”,
In Proceedings of the 2003 ACM symposium on Applied computing
, pp: 582

586,
2003.

[10]

Christos Anagnostopoulos Vassileios Tsetsos, Stathes Hadjiefthymiades, “ On t
he evaluation of semantic web
service matchmaking systems.”,
In Proceedings of the IEEE European Conference on Web Services
, pp: 255

264, 2006.

[11]

Koller D, Sahami M, “Hierarchically classifying documents using very few words”,
In Proceedings of the IEEE
International Conference on Machine

Learning
, pp: 170
-
178, 1997.

[12]

www.w3.org/TR/soap/

[13]

www.w3.org/TR/wsdl

[14]

www.oasis
-
open.org/committees/uddi
-
spec