of 3

Trust for Individuals and Semantic Web Services on the Semantic Web | Semantic Web | Trust (Emotion)

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
ABSTRACT Trust is considered crucial for the success of the Semantic Web. In this paper, research works in this area have been analysed and grouped into two categories, namely 'trust for individuals' also referred to as a 'Web of trust' and 'trust for Semantic Web Services'. Their strategies are compared and analysed. It has been found that the two categories differ in their approach and rarely take costs into consideration.
  Trust for individuals and Semantic Web Services on theSemantic Web Roushdat Elaheebocus School of Electronics and Computer ScienceUniversity of Southampton re1e08@soton.ac.uk ABSTRACT Trust is considered crucial for the success of the Semantic Web. Inthis paper, research works in this area have been analysed andgrouped into two categories, namely 'trust for individuals' alsoreferred to as a 'Web of trust' and 'trust for Semantic WebServices'. Their strategies are compared and analysed. It has beenfound that the two categories differ in their approach and rarelytake costs into consideration. Keywords semantic web, strategies, semantic, trust, provenance 1.INTRODUCTION A Web where people and agents understand one another for better cooperation, such is the vision of the Semantic Web described asan extension of the srcinal Web [1]. Similarly to the real world,for cooperation to happen a certain level of trust is requiredamong the entities involved. The issue of trust is considered socrucial, several researchers have argued that for the Semantic Webto succeed, the issue of trust must be addressed [10, 13, 23].There is no universal definition for trust with respect to thesemantic web but most of them revolve around the idea of aquantifiable level of belief entity A has in entity B, that B iscompetent, will perform reliably to the expectation of A and thelatter accepts to be vulnerable to an acceptable level during a time period and under specific circumstances [2, 3, 4].According to Donovan, A. and Gil, Y. [4], in order to understandthe issue of trust in the context of the Semantic Web, one shouldfirst look at the much bigger picture which goes back to computer science. Research work on trust can therefore be broadlyclassified into four main categories;policy-based, reputation- based, general-models, trust in information resources.Due to space limitation we will focus on the last one that is 'trustin information resources' which in fact makes use of the other three categories and is the most relevant to the Semantic Web.Trust has been given due importance as shown in the semanticcake diagram below: Source: Berners-Lee, T. (2000). Semantic Web onXML.[5] The semantic web cake diagram above shows a dedicated layer for trust at the top which can make use of the different layersunderneath to achieve a semantic web of trust through digitalsignature , encryption and reasoning. As for applications and user interfaces, they are then come on top of the trust layer. Thisdiagram further illustrates the semantic web as an extension of theWeb mentioned earlier. 2.TWO AREAS OF RESEARCH It has been found that there are two main streams of research thathave emerged from the Semantic Web researchers' community:'Trust among individuals' which revolves around the idea of a webof trust [8] and 'Trust for semantic web services' in which userswho can be either humans or intelligent agents making use of services and needing to have a way of determining the trust of services.According to O'Hara, K. et al [15], there are five main strategieswhen tackling the issue of trust on the semantic web namely:optimism, pessimism, centralisation, investigation and transitivityand five type of costs to be taken into consideration; operational,opportunity, risk, deficiency, service payments. We thereforeintend to investigate to what extent O'Hara's observation has beenimplemented. 2.1Trust among Individuals If each web user is to store information about a group of webusers and share it with others, this will result in an effective wayof managing trust on the Web as argued in [8,10] which also  clearly points out that providing security is not equivalent to trust.This approach results in a Web of trust upon which subsequentresearchers have based themselves upon [9, 10, 11, 12, 13]. As aresult, each link between individuals can have trust value [2].In [9], the Friend-Of-A-Friend (FOAF) ontology [14] is extendedwith foaf:person to accommodate a trust value from a scale of 1-9,9 being absolute trust and 1 corresponds to absolute distrust. Theinteresting part is that trust is acknowledged to be not onlydependent on an individual but also the context. Thus, a trustvalue for an entity can vary from context to context. Trust between individuals having no prior interaction is computed usingan algorithm which has been used only for demonstrating theconcept instead of its efficiency. The possible usage of such asystem has been illustrated using a chat-bot which can be queried by passing two email addresses as parameter, and the trust value between them is returned. This works almost in a similar fashionas a web service. And finally, trust value has been embedded in amail client as a new field to indicate the level of trust for eachrecipient.However, [9] has not been tested on a set of real data contrary to[11] where a network of people was considered. In this case trustand distrust have been considered separately. Another major aspect that was neglected in [9] was the time factor.[11] derivesthe importance of time validity from the real world wherebyrelationships are built and maintained with time. Data fromEpinions (Epinions.com) was used, part of which was masked andcross-examined with the system's trust inference to evaluate itsaccuracy.Similarly [10] has also used Epinions in addition to Bibserv toillustrate their trust management system but has assumed that dataon the semantic web is in the form of logical assertions that areconsistent. This leads us to think that it is only when the Web iscompletely 'semantic' that the system will perform optimally.Logical calculus and probabilistic calculus are used to generate atrust value considered to be “a function of his/her (user's) trust inthe sources providing it”The use of the Bayesian decision theory which has its roots in probability is also commonly used in trust evaluation [12, 13].The trust evaluation mechanism proposed in [12] is claimed to beimplementable on any unstructured peer-to-peer network. Basedupon reputation gathered through interaction with peers, theinformation can then be shared over the network to produce aglobal rating. This approach has been evaluated throughsimulations only unlike [9, 10].According to [13], trust is a probabilistic interpretation andtherefore to have a higher accuracy, the combination of information from a variety of sources is performed.We find that when trust is considered for individuals, the'transitivity' strategy [15] is used mostly. One aspect which has been neglected in most of the research work above however is thecost except for [13] in which three kinds of cost were taken intoaccount namely: operational, opportunity, service payments. It isalso important to note that all the works cited above have claimedto provide personalised trust. 2.2Trust for Semantic Web Services Semantic web services are accessible not only by people but alsothrough agents and for that purpose, semantic markup languagesfrom the DAML family were introduced [16]. However enablingusage by agents is not enough and requires that the latter findways of determining whether a service can be trusted and if yes, towhat extent.Modelling elements were added to the Web Service ModellingOntology (WSMO) in [20] to include information about trust.Service providers do not publish their services to a third partyregistry but instead, join a peer-to-peer network of registriesthemselves when they want to provide a service. This effectivelyaddresses the limitation caused by centralised matchmaking andallows providers to retain control over their policies. Thelanguage used in this system is known as PeerTrust described as alanguage for trust negotiation that supports delegation, policy protection and negotiation strategies [21] and is built upon therule layer of the semantic web cake [5].Another language used is the Poof Markup Language (PML) in[22] to provide justification for the result provided by a webservice. This is particularly important to enable agents to reason.Expressed in OWL, PML is thus compatible with semantic webservices and clients. One example of its usage is demonstrated inIWTrust [18] where it is used for providing additional informationabout provenance on answers obtained from the web enabling a better evaluation of trustworthiness.Provenance is of specially high concern [17, 18, 19] when dealingwith web services . [19] describes a trust aware inferenceframework and an ontology to represent associations of trust and provenance. The overall system is in the form of a semantic webservice that “evaluates trustworthiness of semantic associationfrom multiple sources”.We see a difference between trust in individuals where thetransitive approach is favoured compared to semantic webservices whereby a more investigative approach [15] is used since provenance decreases the amount of uncertainty in a source. Thecosts of such strategies are not really discussed in most of theresearch considered for semantic web services. 3.CONCLUSION Trust is indeed a crucial issue for the success of the semantic web.There exists multiple heterogeneous strategies for tackling trust onthe semantic web that are often not inter-operable which is notnecessarily completely a bad thing since each may be appropriatein specific contexts. As these strategies become more mature,standards will emerge and the issue of interoperability will besolved. However, research works tend to ignore the different costsdescribed in [15] which can result in some undesired effect whendeployed that may not be apparent in simulation or throughsample data.Also, trust srcinates from social behaviour and most solutions for trust on the semantic web take a solely technological approach.External disciplines such as sociology, law and other cognitivesciences should be considered as well to have a better understanding of trust and how to address it [6]. Web Science [7]does exactly this and hopefully better solutions will come out of this in the near future.  4.REFERENCES [1]Berners-Lee, T., Hendler, J. and Lassila, O. “The SemanticWeb,” Scientific Am. , May 2001, pp. 34–43.[2]Grandison, T. and Sloman, M. A survey of trust in Internetapplications. IEEE Communications Surveys and Tutorials, 4(2000):2–16.[3]Olmedilla, D., Rana, O., Matthews, B., and Nejdl, W.Security and trust issues in semantic grids. In Proceedings of the Dagsthul Seminar, Semantic Grid: The Convergence of Technologies,(2005) volume 05271.[4]Donovan,A. and Gil, Y. “A survey of trust in computer science and the Semantic Web.” Web Semant. 5, no. 2(2007): 58-71.[5]Berners-Lee, T. Semantic Web on XML. Presentation atXML 2000, available fromhttp://www.w3.org/2000/Talks/1206-xml2k-tbl/slide10-0.html. Accessed 14 May 2009.[6]O'Hara, K. and Hall, W., “Trust on the Web: Some WebScience Research Challenged in “Web Science” Dossier. Iss.7, October 2008.http://www.uoc.edu/uocpapers/7/dt/eng/webscience.pdf Accessed 10 May 2009.[7]“MIT and University of Southampton launch World WideWeb research collaboration - MIT News Office.”http://web.mit.edu/newsoffice/2006/wsri.html. Accessed 10May 2009.[8]Rohit, K. and Rifkin, A. “Weaving a Web of trust.” WorldWide Web J. 2, no. 3 (1997): 77-112.[9]Golbeck,J., Parsia, B. and Hendler, J. “Trust Networks onthe Semantic Web”, In Proceedings of CooperativeInformation Agents VII, 2003, 238-249.[10]Matthew,R., Agrawal,R. and Domingos, P. “TrustManagement for the Semantic Web.” In The SemanticWeb -ISWC 2003, 351-368, 2003.[11]Guha, R., Kumar, R., Raghavan, P. and Tomkins, A..“Propagation of trust and distrust.” In Proceedings of the13th international conference on World Wide Web, 403-412. New York, NY, USA: ACM, 2004.[12]Wei,W., Zeng,G. and Yuan, L. “A Semantic ReputationMechanism in P2P Semantic Web.” In The Semantic Web – ASWC 2006, 682-688, 2006.[13]Xiaoqing, Z., Chen, H., Wu, Z., and Zhang, Y. “AComputational Trust Model for Semantic Web Based onBayesian Decision Theory.” In Frontiers of WWW Researchand Development - APWeb 2006, 745-750, 2006.[14]Dumbill, Ed, “XML Watch: Finding friends with XML andRDF.” IBM Developer Works, http://www-106.ibm.com/developerworks/xml/library/xfoaf.html , June2002. Accessed 14 May 2009.[15]O'Hara, K., Alani, H., Kalfoglou,Y. And Shadbolt, N. “TrustStrategies for the Semantic Web”, in Proceedings of theISWC2004 Workshop on Trust, Security, Reputation on theSemantic Web, 2004.[16]McIlraith, S., Son, T., and Zeng.H., “Semantic webservices”. IEEE Intelligent Systems, Special Issue on theSemantic Web, 16(2):46/53, March/April 2001.[17]Carroll, J., Bizer, C., Hayes,P. and Stickler, P.. “Namedgraphs, provenance and trust.” In Proceedings of the 14thinternational conference on World Wide Web, 613-622.Chiba, Japan: ACM, 2005.[18]Ilya,Z., Pinheiro da Silva,P. and L. McGuinness, D..“IWTrust: Improving User Trust in Answers from the Web.”In Trust Management, 384-392, 2005.[19]Ding,L., Kolari, P., Finin, T., Joshi, A., Peng,Y. and Yesha,Y. “On Homeland Security and the Semantic Web:.” INAAAI SPRING SYMPOSIUM ON AI TECHNOLOGIESFOR HOMELAND SECURITY (2005): 21--23.[20]Olmedilla, D., Rubén, L., Axel, P., and Lausen, H. “Trust Negotiation for Semantic Web Services.” In Semantic WebServices and Web Process Composition, 81-95, 2005.[21]Wolfgang, N., Olmedilla,D. and Winslett, M. “PeerTrust:Automated Trust Negotiation for Peers on the SemanticWeb.” In Secure Data Management, 118-132, 2004.[22]Paulo Pinheiro da, S., L. McGuinness, D. and Fikes, R. “A proof markup language for semantic web services.” Inf. Syst.31, no. 4 (2006): 381-395.[23]Tim, F. and Joshi, A. “Agents, trust, and information accesson the semantic web.” SIGMOD Rec. 31, no. 4 (2002): 30-35.
Related Search
Related Docs
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks