See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/257418229

SACoSS-semantic Agent Based System for

Cloud Service Suggestion Using Cloud Service
Ontology
ARTICLE in JOURNAL OF COMPUTATIONAL INFORMATION SYSTEMS NOVEMBER 2012

DOWNLOADS

VIEWS

242

84

3 AUTHORS, INCLUDING:
Saravana Balaji B
S.Veerasamy Chettiar College of Engineerin
10 PUBLICATIONS 0 CITATIONS
SEE PROFILE

Available from: Saravana Balaji B

Retrieved on: 12 August 2015

Journal of Computational Information Systems 8: 21 (2012) 87638773

Available at http://www.Jofcis.com

SACoSS-semantic Agent Based System for Cloud Service

Suggestion Using Cloud Service Ontology
B.Saravana Balaji 1,,
1 Research
2 Assistant
3 Professor

T.N.Prabhu 2 , Dr.N.K.Karthikeyan 3

Scholar, Anna University of Technology, Coimbatore, India

Professor-IT, Sri Ramakrishna Engineering College, Coimbatore, India

and Head-IT, Sri Krishna College of Engineering and Technology, Coimbatore, India

Abstract
With increasing demand and adoption of cloud services through web, selecting an appropriate cloud
service according to requirements of the consumer becomes a great issue. In this paper we present a
system for providing a list of cloud services as suggestion according to the consumer requirement, called
SACoSS which is built using semantic technologies. SACoSS divides the cloud services in two domains
based on SLA and WSDL of cloud services as SaaS services and IaaS services and updates the ontologies
accordingly. The workow of SACoSS system is of two steps, cloud service ontology updation and cloud
service suggestion generation. The performance evaluation of SACoSS shows that this system provides
better results and achieves higher prediction accuracy than other methods.
Keywords: Cloud Computing; Ontology; SLA; Semantic Agents; Service Selection

Introduction

The emergence of Grid Computing and Service Oriented Architecture makes a new computing
paradigm called cloud computing. The main idea behind the emergence of cloud computing is
to reduce the cost of computing and also to increase the reliability, availability and exibility [8].
Cloud computing is provided in three formats called Software as a Service (SaaS), Platform as
a Service (PaaS), Infrastructure as a Service (IaaS). The main players of cloud computing are
cloud service providers and cloud service consumers. In cloud computing, the requirement of
service consumer is to have the minimum hardware and software, and need only pay to service
providers to get the required compute resources. Under the diverse nature of cloud computing,
a cloud service consumer faces numerous technical issues on selecting a particular cloud service
like, the consumer must understand the technical requirements, and workow of the service
etc. Normally, these kinds of information about a particular cloud service are provided through
Service Level Agreements (SLA)and WSDL documents by the cloud service provider. SLAs
contains information like trading parties, attributes, measurement metrics, and attribute values

Corresponding author.
Email address: saravanabalaji.b@gmail.com (B.Saravana Balaji).

15539105 / Copyright
November 1, 2012

2012 Binary Information Press

8764

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

etc [19]. Regardless of availability of SLAs and WSDL documents the cloud service consumer
faces the problem of selecting the appropriate service according to the need as the SLAs describe
the services through non-standardized attributes and WSDL document describes service technical
attributes [18]. To solve this issue, semantic web technologies could be used.
In the Semantic Web, knowledge is represented by means of ontologies [3], which are viewed
as a formal specication of a domain knowledge conceptualization. The semantic descriptions of
services in the form of ontologies have emerged due to their ability for supporting the automation of service retrieval, invocation, composition and monitoring tasks by providing machineexploitable meaningful declarative descriptions of service characteristics [5]. In this paper, we
propose SACoSS-Semantic Agent based System for Cloud Service Suggestion using Cloud Service
Ontology, which uses the SLA and WSDL document of cloud services to extract the knowledge
about the cloud service and answers the cloud service requester (consumer) with appropriate
cloud services as suggestion according to the requirements of the requester (consumer).
The remainder of this paper is organized as follows. Section II describes various related methods
proposed previously. Section III explains about the architectural components of the SACoSS.
Section IV provides information about workow of SACoSS. Section V gives details about the
performance evaluation of the SACoSS by comparing with few existing methods under dierent
metrics. Section VI consists of conclusion and future work that can be extended from the SACoSS.

Related Work

The problem of selecting an appropriate service involves mutlicriteria decision making situation
and there has been several methods were proposed by many researchers around the globe. The
method proposed in [4] addresses Automated web-based negotiation of Service Level Agreements
(SLA) which can help dene the QoS requirements of critical service-based processes, is achieved
through a novel trusted Negotiation Broker (NB) framework and by dening mathematical models. Also, intelligent selection algorithm is being used for adopting decision functions. A method
for the construction of ontology automatically is proposed in [23] by mining large-scale web resources such as eHow and wikiHow. This method attempts to build a huge situation knowledge
base of human activities by means of text mining techniques that exploit the structure of the
how-to descriptions, which is essential for context/situation-aware services.
A recommendation system for home network environment has been proposed in [11]. This
method constructs ontologies for user proles and home network service and by using those, a
recommendation system has been developed. For resource selection in science cloud, a method has
been proposed in [9]. This method describes an ontology-based representation of cloud computing
environment which allows conceptualization of common attributes among cloud resources and
describes relations among them semantically. A resource virtualization method using ontology is
introduced and a new Virtual Ontology (VOn) is congured dynamically. For rapid and ecient
merging of a number of ontologies, the Map Reduce model is adopted.
Web service selection based on context ontology and QoS has been proposed in [16]. It describes
how automatic selection of best service provider done upon a request based on mixed context and
QoS ontology for a given set of parameters of QoS. It supports an e-business framework and also
emphasizes one dimensional quality of services. Discovering semantic web service using SPARQL
and agents has been proposed in [15]. It describes how the SPARQL query language can be used

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

8765

to express the preconditions and postconditions of services as well as the goals of agents. Here
SPARQL query evaluation can be used to check the truth of a precondition in a given context,
construct the postcondition that will result from the execution of a service in a context, and also
determine whether a service execution with those results will satisfy the goals of an agent.
A service concept recommendation system for service ecosystem has been proposed in [6]. It describes the framework of service concept recommendation system, which is built upon a semantic
similarity model. This method did ontology-based service matchmaking from the perspective of
measuring the similarity of concept-concept, instead of measuring the similarity of query-concept
utilized by those semantic matchmakers. In [24] a collaborative ltering approach for predicting
QoS values of Web services and makes Web service recommendation by implementing a prototype
called WSRec which helps to achieve better prediction accuracy has been proposed. In [21] an
ontology based network management system has been proposed. It presents a system to collect
information through the cooperation of intelligent agent software, in addition to providing warnings after analysis to monitor and predict some possible error indications among controlled objects
in the network. This system could sketch the four main components of network management systems with the technique of graphic monitoring multi-agent, which eectively enhance and improve
the network monitoring performance. A semantic based automated service discovery method has
been proposed in [1]. It addresses the issue of Web Service discovery given non-explicit service
description semantics that match a specic service request by semantic based service categorization and semantic enhancement of the service request and provides a solution for functional level
service categorization. Latent Semantic Indexing (LSI) is utilized for ecient matching.

SACoSS - Architecture

SACoSS architecture is provided in gure 1. SACoSS consists of seven semantic agents and a
cloud service ontology repository for suggesting appropriate cloud services.

Fig. 1: SACoSS - Architecture

The semantic agents of SACoSS system are
SLA Processing Agent

8766

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

Semantic Annotating Agent

Domain Specifying Agent
SaaS Ontology Updating Agent
IaaS Ontology Updating Agent
Query Processing Agent
Cloud Service Searching Agent

3.1

SLA processing agent

This agent is responsible for retrieving SLA and WSDL documents from various cloud services
that are available on the web. Then this agent process the SLA and WSDL documents based on
PoS tagging method dened in RASP system [2] through which the environmental template of
the cloud service developed.

3.2

Semantic annotating agent

This agent takes input from SLA processing agent and adds annotations for the subjecs or nouns
present on the parsed documents. This annotation process is based on WSDL-S [12].

3.3

Domain specifying agent

This agent identies the domain of the cloud service that is descried in the environmental template
of the cloud service. Depending up on the domain of the cloud service the template will be
forwarded to either SaaS ontology updating agent or IaaS ontology updating agent.

3.4

SaaS ontology updating agent

This agent updates the existing ontology for Software as a Service kind cloud services by adding
classes and object values for that service via the information that is described in the environmental template and store it in the cloud service ontology repository. The ontology structure and
processing method for updating ontology will be discussed in the next section.

3.5

IaaS ontology updating agent

This agent updates the existing ontology for Infrastructure as a Service kind cloud services by
adding classes and object values, adding relationship between various resources of the service with
the help of information that is described in the cloud service template and store it in the cloud
service ontology repository. The ontology structure and processing method for updating ontology
will be discussed in the next section.

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

3.6

8767

Query processing agent

This agent will parse the query given by the user in natural language by using the PoS tagging
method dened in RASP system [2] and develops the operational requirement of the cloud service
requester by using the method dened in [13].

3.7

Cloud service searching agent

This agent will communicate the domain specifying agent and forwards the requirement of the
cloud service consumer. The domain specifying agent will identify the domain of the particular
requirement and send back the result. The cloud service searching agent will generate SPARQL
query according to the requirement and search the CS ontology repository for the services and
provide the list of cloud services as suggestions to the cloud service consumer. The query language
that is used by this system is SPARQL [17].

Workflow of SACoSS

The workow of SACoSS system can be divided into two divisions, Cloud Service Ontology
Updation and Cloud Service Suggestion Generation. SACoSS system is capable of dierentiate
cloud services in terms of SaaS and IaaS only, where the constraints for dening a cloud service
as PaaS service are not considered during the development of the system.

4.1

Cloud service ontology updation

The SLA processing agent has a sub agent called service crawler whose duty is to retrieves the SLA
and WSDL document of the Cloud Service Providers from the web. The SLA processing agent
process the documents through RASP system [2] and generates the environmental template of the
cloud service by the natural language processing method dened in [13] where the technical and
non functional attributes of cloud services are predened and extracted, for this purpose around
320 attributes type have been identied and used. Then the service template is passed through
Semantic Annotating agent which adds annotation to the words of the document by means of [12].
Then the template will be passed on to domain specifying agent. The domain specifying agent
then identies the domain of the particular service by matching the words of service template
with predened keywords of SaaS and IaaS services like software, storage, uptime of service etc.
For matching purpose around 140 keywords has been identied to segregate a cloud service as
IaaS or SaaS cloud service. The matching technique follows a simple rule - majority of keyword
matches then the service domain is that.
Depend upon the service type the domain specifying agent then forwards the service template
to the corresponding ontology updating agent (i.e. SaaS or IaaS). Both the ontology updating
agents follows the updating process to update the cloud service ontology, where the cloud service
ontology structure is developed based on OWL-S [22]. To reduce the complexity, without non
functional attributes, the reduced cloud service ontology structure for SaaS services is given in
gure 2. SaaS cloud service ontology contains classes for representing the software details like
browser support, minimum hardware requirement, user credentials service type, usage limit etc.,

8768

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

along with non functional attributes.

Fig. 2: SaaS ontology structure

The reduced cloud service ontology structure for IaaS services without non functional attributes
is given in gure 3. IaaS cloud service ontology consists of classes for representing infrastructure
details like software and hardware requirements, service type, user credentials, storage space limit,
availability of the service, integrity constraint, encryption option, backup option etc., with non
functional attributes. The ontology updating agents parse the cloud service templates and update
the corresponding service attributes as properties of the ontology classes.

4.2

Cloud service suggestion generation

The SACoSS system takes query from the user in natural language. The system provides an
option for adding annotation for the query; if the cloud service consumer is an expert user then
the consumer can add annotation for the query. For example if consumer is in need of 5GB
of storage space then consumer can provide annotation for the word storage like memory, raw
disk etc. The query processing agent follows the same mechanism used by SLA processing agent
and develops the query structure. Then the query is sent to the domain specifying agent which
nds the domain of the query and forwards the query with appropriate domain to cloud service
searching agent.
The cloud service searching agent generates the SPARQL query according to the query structure
and submit the query to CS-ontology repository which in turn returns the appropriate results.
The algorithm used to generate query and to retrieve result is given below
(1) For each request
a. Develop the object restrictions
b. Develop the restriction for datatype properties
(2) Develop the query
a. Input - set of objects with value attributes
b. Create select query for domain class
(1) If there is a condition

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

8769

Fig. 3: IaaS ontology structure

(2) Create Filter sub query for value attributes
(3) If there is a combination for service conditions
(4) Create Union sub query for value attributes
(3) Find appropriate ontology type depending on domain class of query
(4) Submit query and retrieve result
(5) Generate list of service in table format from result
The cloud service searching agent produces the services list in table format to the cloud service
consumer.

Performance Evaluation of SACoSS

SACoSS system has been implemented using Jena semantic web framework [10] and SPARQL for
querying the cloud service ontology repository through ARQ. The cloud service ontology repository is a MySQL database which is accessed through Jena for the storage of service ontology.
Apache Tomcat was used to host the CS-ontology repository. The agents are given access through
a JAVA API which has the functionality of querying. The list of services used by SACoSS is derived from [7] which provide details about cloud service providers under dierent categories. For
SaaS service ontology, 15 domains have been identied and 178 services are classied according to their domains. For IaaS service ontology, 6 domains have been identied and 94 services
are stored accordingly. To evaluate the performance of SACoSS, it has been tested in a Pentium Xeon, RedHat Linux server system and uses Semantic Web Service Matchmaker Evaluation
Environment[20] for comparing the results with [16, 1] and OWLS-M4 [14]. The Semantic Web
Service Matchmaker Evaluation Environment evaluates semantic service selection systems over
given test collections in terms of standard retrieval performance evaluation measures and provides
output through chart, graph, XML. The measures used to evaluate the SACoSS performance are
query response time, precision recall and fallout. Figure 4 depicts precision recall curves comparison and gure 5 depicts fallout recall curves comparison. For comparing query response time,
20 queries have been given as input to all the four systems. The complexity of queries increases

8770

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

for each one, the number of words and requirements for a service increases from query to query.
Figure 6 depicts the comparison of query response time of all the four methods.The results reveals
that SACoSS system outperforms other three methods in most of the cases and performs in par
with them in few cases.

Fig. 4: Precision-recall curve

Fig. 5: Fallout-recall curve

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

8771

Conclusion and Future Work

For a naive user of cloud computing, to select an appropriate cloud service is a dicult task
as it involves understanding of SLA documents and WSDL documents which provides technical
and non functional aspects of a cloud service. Hence to overcome this issue we presented a
system called SACoSS for cloud service suggestion according to the requirements of the cloud
service consumer by using cloud service ontology. SACoSS has been developed using various
semantic technologies. This system comprises seven semantic agents and a cloud service ontology
repository. The workow of SACoSS can be divided in to two sections, cloud service ontology
updation and cloud service suggestion generation. SACoSS system is capable of suggesting SaaS
kind cloud service and IaaS kind cloud services. The performance evaluation of the SACoSS
system reveals that it performs well than other similar kind of methods that has been proposed
earlier.

Fig. 6: Query response time

Fig. 7: Abbreviations

In the view of extending SACoSS system, an ontology structure for PaaS kind of services can be
developed and incorporated with it. SACoSS system does not provide any ranking to the services
and a new ranking methodology for services can be developed and incorporated with SACoSS.
While selecting a service, taking user reviews about that service in to account as a constraint
could be a good practice. Hence developing and incorporating user reviews with each service in
service ontology could become a good extension for SACoSS.

8772

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

References
[1]

Aabhas V. Paliwal, Basit Shaq, Jaideep Vaidya, Hui Xiong, Nabil Adam, Semantics Based Automated Service Discovery, IEEE Transactions on Services Computing, April 2011.

[2]

Briscoe, E., J. Carroll and R. Watson, The Second Release of the RASP System, In Proceedings
of the COLING/ACL 2006 Interactive Presentation Sessions, Sydney, Australia.

[3]

Dieter Fensel and Frank van Harmelen, Vrije Universiteit, Amsterdam, An Ontology Infrastructure
for the Semantic Web, IEEE Transactions on Knowledge and Data Engineering, 2001, Vol. 19 No.
2 pp. 261-272.

[4]

Farhana H. Zulkernine and Patrick Martin, An Adaptive and Intelligent SLA Negotiation System
for Web Services, IEEE Transaction on Services Computing, January-March 2011, pp: vol. 4, no.
1.

[5]

George A. Vouros, et al. A semantic information system for services and traded resources in Grid
e-markets, Future Generation Computer Systems, 2010.

[6]

Hai Dong, Farookh Khadeer Hussain, ElizabethChang, A service concept recommendation system for enhancing the dependability of semantic service matchmakers in the service ecosystem
environment, Journal of Network and Computer Applications 34 (2011), pp: 619-631.

[7]

http://www.cloudbook.net/directories/product-services/cloud-computing-directory

[8]

B. Hayes, Cloud computing, Communications of the ACM 51 (7), 2008.

[9]

Hyunjeong Yoo, Cinyoung Hur, Seoyoung Kim, and Yoonhee Kim, An Ontology-based Resource
Selection Service on Science Cloud, International Journal of Grid and Distributed Computing,
December, 2009, pp: vol. 2, no. 4.

[10] Jena A Semantic Web Framework for Java, http://incubator.apache.org/jena/.

[11] Jinhyung Kim, Dongwon Jeong, Doo-Kwon Baik, Ontology-based Semantic Recommendation System in Home Network Environment, IEEE Transactions on Consumer Electronics, August 2009,
pp: vol. 55, no. 3.
[12] John Miller, et al. WSDL-S: Adding Semantics to WSDL, white paper, University of Georgia.
[13] Jong Hyun Lim, Kyong-Ho Lee, Constructing composite web services from natural language requests, Web Semantics: Science, Services and Agents on the World Wide Web, 2010, pp. 1-13.
[14] M. Klusch, B. Fries, K. P. Sycara, OWLS-MX: a hybrid Semantic Web service matchmaker for
OWL-S services, Journal of Web Semantics 7 (2) (2009), 121-133.
[15] Marco Luca Sbodioa, David Martinb, Claude Moulin, Discovering Semantic Web services using
SPARQL and intelligent agents, Web Semantics: Science, Services and Agents on the World Wide
Web 8 (2010), pp: 310-328.
[16] Nabil Keskes, Ahmed Lehireche and Abdellatif Rahmoun, Web Services Selection Based on Context Ontology and Quality of Services, International Arab Journal of e-Technology, January 2010,
pp: vol. 1, no. 3.
[17] Prudhommeaux, E., and Seaborne, A. (2008). SPARQL query language for RDF, W3C recommendation, W3C, available at http://www.w3.org/TR/2008/REC-rdf-sparql-query-20080115.
[18] M. Risch, J. Altmann, Enabling Open Cloud Markets through WS-Agreement Extensions, Service
Level Agreements in Grids Workshop, in conjunction with GRID 2009, in: CoreGRID Springer
Series, Ban, Canada, 2009.
[19] M. Risch, I. Brandic, J. Altmann, Using SLA mapping to increase market liquidity, NFPSLAMSOC 2009, in conjunction with the 7th international joint conference on service oriented computing,
Stockholm, Sweden, November, 2009.

B. Balaji et al. /Journal of Computational Information Systems 8: 21 (2012) 87638773

8773

[20] Semantic Web Service Matchmaker Evaluation Environment (2009), http://projects. semwebcentral.org/projects/sme2/.
[21] Sheng-Yuan Yang, Yi-Yen Chang, An active and intelligent network management system with
ontology-based and multi-agent techniques, Expert Systems with Applications 38 (2011), pp:
10320-10342.
[22] World Wide Web Consortium, OWL-S: Semantic Markup for Web Services, World Wide Web
Consortium (W3C) Member Submission, 2004, available at http://www.w3.org/submission/owl-s.
[23] Yuchul Jung, Jihee Ryu, Kyung-min Kim, Sung-Hyon Myaeng, Automatic construction of a largescale situation ontology by mining how-to instructions from the web, Web Semantics: Science,
Services and Agents on the World Wide Web 8 (2010), pp: 110-124.
[24] Zibin Zheng, Hao Ma, Michael R. Lyu and Irwin King, QoS-Aware Web Service Recommendation
by Collaborative Filtering, IEEE Transaction on Service Computing, April-June 2011, pp: vol. 4,
no. 2.