Trust on the Semantic Web

1. Trust on the Semantic Web Seyyed asgary ghasempouri Sharif University of Technology Computer Department

2. Outline • Web of Trust? • Objective of paper & Contributions • Networks in Semantic Web? • How to build a Trust Network? • Trust Graph • Computation of Trust • Trust Web Service • Applications -> TrustBot, TrustMail • Related Works • Conclusion Sharif University of Technology Computer Department

3. Web of Trust? • Web of trust-> each user explicitly specify a (possibly small) set of users she trusts. The resulting web of trust may be used recursively to compute a user’s trust in any other user • Web of trust • Research has been concentrated more on source of information which misses trust in terms of human sense. • Focused largely on digital signatures, certificates, authentications. Sharif University of Technology Computer Department

4. Contributions • Apply social networks to semantic web • Consider trust in to account with a much more human sense. • Ex: How much credence should I give to a what this person says about a topic • The degree of trust associated with it could be based on your past encounter or could be based on what your friends says about him • Build a Trust Network extending FOAF ontology & by adding their own Trust Ontology • Compute trust values between two people • Illustrate its usefulness using applications Sharif University of Technology Computer Department

5. Networks on Semantic Web • Information is machine readable • Concepts in semantically marked up pages are automatically linked through ontological relations • visualized as a large graph where web resources are nodes & edges form relations between objects or webpages • Generating Social Networks • Individuals manage data about themselves and their friends • Information about individuals in a network is maintained in distributed sources • Digital signature can be associated to files going across the network • Security measures builds trust about the authenticity or data contained within the network Sharif University of Technology Computer Department

6. Building Trust Network • FOAF can be used to describe information about himself, such as name, email address, homepage, people he knows • Extended FOAF ontology (Friend-Of-A-Friend) • Following properties were added to it, which allows users to indicate a level of trust for people they know • Trust properties • Trusts neutrally, Trusts slightly, Trusts moderately, Trusts highly, Trusts absolutely • Distrust properties • Distrust absolutely, Distrust highly, Distrust moderately, Distrust slightly • Users can sign these files so that information source can be verified Sharif University of Technology Computer Department

7. Example 1 <Person rdf:ID="Joe"> <mbox rdf:resource="mailto:bob@example.com"/> <trustsHighly rdf:resource="#Sue"/> </Person> Sharif University of Technology Computer Department

8. Example 2 <Person rdf:ID="Bob"> <mbox rdf:resource="mailto:joe@example.com"/> <trustsHighlyRe> <TrustsRegarding> <trustsPerson rdf:resource="#Dan"/> <trustsOnSubject rdf:resource="http://example.com/ont#Research"/> </TrustsRegarding> </trustsHighlyRe> <distrustsAbsolutelyRe> <TrustsRegarding> <trustsPerson rdf:resource="#Dan"/> <trustsOnSubject rdf:resource="http://example.com/ont#AutoRepair"/> </TrustsRegarding> </distrustsAbsolutelyRe> </Person> Sharif University of Technology Computer Department

9. Trust Graph • Directed Edges in the graph contain explicitly specified trust values • It can be used to infer the trust values between two people who are not directly connected • Several Basic calculations • Maximum and minimum capacity paths • Identify the trust capacity of the paths with highest lowest respectively • Determined by making a network flow calculation for each individual path between the source and sink • Maximum amount of trust a source can give to a sink is limited by the smallest edge weight along the path Sharif University of Technology Computer Department

10. Trust Graph Contd.. • Maximum and minimum length paths • measure of the number of edges between the source and the sink • Weighted average between two people (node X & Y) • General notion is that users would want lower trust ratings for someone many links away as opposed to a direct neighbor • Distrust notion is very ambiguous: • Ex: A distrust B regarding a specific subject and in turn, B distrust C on that subject, it is possible that A distrust C, or A trust C. Sharif University of Technology Computer Department

11. Trust Calculation • It uses the maximum capacity of each path to the sink • Algorithm is recursive & calculates the average • For any node that has direct edge to sink node , they ignore the paths & use the direct edge weight. • Otherwise they determine the weighted average values for each of the neighbors, which have a path to sink Sharif University of Technology Computer Department

12. Trust Web Service • Trust Web service • A web users can provide two email addresses & in return the service would return the weighted average • User can provide their own algorithms for calculating trust • It retrieves the neighbors, gets the list of trust rating for a given edge, detecting the presence or absence of path between two individuals, & finding path lengths. Sharif University of Technology Computer Department

13. Applications -TrustBot • TrustBot is an IRC bot. • Gives trust recommendations when call is made • Builds an internal representation of the trust network from a collection of distributed sources. • User can query from IRC channel, & the bot returns the trust values • Provides the weighted average, as well as maximum and minimum path lengths, and maximum and minimum capacity paths Sharif University of Technology Computer Department

14. Applications - TrustMail • Email client, developed on top of Mozilla Messenger • provides an inline trust rating for each email message • calls the web service, passing in the email address of the sender & mailbox address • If a user has a trust rating with respect to email, that value is used else general trust rating is used Sharif University of Technology Computer Department

15. TrustMail Sharif University of Technology Computer Department

16. TrustMail Contd.. Scenario • Two groups of people • Each group has a Professor & set of students • The two professors know each other & have their trust ratings in trust Graph • “My advisor has collaborated with you on this topic in the past and she suggested I contact you.” • Professor on receiving the email needs to verify either by calling the other professor etc.. • Using TrustMail reduces this by providing trust ratings for each emails & may be with respect to the email subject topic • Their Claim • TrustMail lowers the cost of sharing trust judgments across widely dispersed and rarely interacting groups of people Sharif University of Technology Computer Department

17. Related Works • Social Network & application of “small world” notion • ”Small World” notion by Stanley Milgram, almost everybody in the world are at most separated by “six degrees of separation” • Complex networks show this “small world” phenomenon • Small average distance between two nodes, a high connectance or clustering co-efficient • “Smallworld” have been studied with respect to random graphs. Studies have been undertaken with respect to spread of diseases between networks Sharif University of Technology Computer Department

18. Related Works -Trust on the Semantic Web • Yolanda Gil and Varun Ratnakar • Addressed trusting content and information sources • Users included the credibility and reliability values while annotating • Their trust assessments were based on individual feedback about the source of information • Trust values are averaged and presented to the viewer. • Uses TRELLIS system, users could view information, annotations (averages of credibility, reliability etc) and then make analysis. Sharif University of Technology Computer Department

19. Conclusions • Social Networks exists in the current web • In current web, its hard to determine the topic based on which the clustering (or social networks have been formed) • In Semantic Web everything is machine readable, & trust information can be annotated along with FOAF, so that trust can be associated with individuals in social networks Sharif University of Technology Computer Department

20. Conclusions • Trust network is a directed graph with nodes forming the person and edges forming the trust weights • Trust value computed is based on the following • Priority is given to direct link between two people • Otherwise they try to find a weighted average of the path between X & Y. Sharif University of Technology Computer Department

21. Conclusion • Concept of trust and distrust is subjective, there can be several different metrics for inferring trust values between two people • Authors, do not concentrate of developing an optimal algorithm for computing trust • Authors focus on simple algorithm • They try show some applications in which trust ratings can be used- TrustMail Sharif University of Technology Computer Department

22. Conclusion • Good thing about the paper is that they build trust networks on semantic web in a much more human sense. • They show that some of the applications like TrustMail can utilize the trust ratings. • Their claim is that Trust values can be inferred between two people even though there isn’t direct trust rating. Sharif University of Technology Computer Department

23. References • Jennifer Golbeck link to Web of Trust, http://trust.mindswap.org/cgi-bin/trustBuilder.cgi • Trust Networks on the Semantic Web -Jennifer Golbeck, Bijan Parsia, James Hendler Sharif University of Technology Computer Department

24. References • 1. Adamic, L., "The Small World Web". Proceedings of ECDL, pages 443-- 452, • 1999. • 2. Adding SVG Paths to Co-Depiction RDF, • http://Jibbering.com/svg/codepiction.html • 3. The Advogato Website: http://www.advogato.org • 4. Albert, R., Jeong, H. AND Barabasi, A.-L. "Diameter of the world-wide web." • Nature 401, 130–131, 1999 • 5. Bharat, K and M.R. Henzinger. "Improved algorithms for topic distillation in a • hyperlinked environment," Proc. ACM SIGIR, 1998. • 6. Brin, S and L. Page, "The anatomy of a large-scale hypertextual Web search • engine," Proc. 7th WWW Conf., 1998. • 7. Broder, R Kumar, F. Maghoul, P. Raghavan, S. Rajagopalan, R. Stata, A. • Tomkins, and J. Wiener. "Graph structure in the web. " Proc. 9th International • World Wide Web Conference, 2000. • 8. Carriere, J and R. Kazman, "WebQuery: Searching and visualizing the Web • through connectivity," Proc. 6th WWW Conf., 1997. • 9. Chakrabarti, S, B. Dom, D. Gibson, J. Kleinberg, P. Raghavan, and S. • Rajagopalan, "Automatic resource compilation by analyzing hyperlink structure • and associated text," Proc. 7th WWW Conf., 1998. Sharif University of Technology Computer Department

25. References • 10. Dumbill, Ed, “XML Watch: Finding friends with XML and RDF.” IBM • Developer Works, http://www-106.ibm.com/developerworks/xml/library/xfoaf. • html, June 2002. 11. FOAFNaut: http://foafnaut.org/ • 12. Gil, Yolanda and Varun Ratnakar, "Trusting Information Sources One Citizen at • a Time," Proceedings of the First International Semantic Web Conference • (ISWC), Sardinia, Italy, June 2002. • 13. Kleczkowski, A. and Grenfell, B. T. "Mean-fieldtype equations for spread of • epidemics: The ‘small-world’ model." Physica A 274, 355–360, 1999. • 14. Kleinberg, J, "Authoritative sources in a hyperlinked environment," Journal of • the ACM, 1999. • 15. Kumar, Ravi, Prabhakar Raghavan, Sridhar Rajagopalan, D. Sivakumar, Andrew • Tomkins, and Eli Upfal. "The web as a graph". Proceedings of the Nineteenth • ACM SIGMOD-SIGACT-SIGART Symposium on Principles of Database • Systems, May 15-17, 2000. • 16. Labalme, Fen, Kevin Burton, "Enhancing the Internet with Reputations: An • Openprivacy Whitepaper," http://www.openprivacy.org/papers/200103- • white.html, March 2001. • 17. Levien, Raph and Alexander Aiken. "Attack resistant trust metrics for public key • certification." 7th USENIX Security Symposium, San Antonio, Texas, January • 1998. Sharif University of Technology Computer Department

26. References • 18. Milgram, S. "The small world problem." Psychology Today 2, 60–67, 1967. • 19. Moore, C. and Newman, M. E. J. "Epidemics and percolation in small-world • 20. Newman, Mark, "The structure of scientific collaboration networks," Proc. Natl. • Acad. Sci. USA 98, 404-409 (2001). • 21. Newman, Mark, "Models of the small world", J. Stat. Phys. 101, 819-841 (2000). • 22. Open Privacy Initiative: http://www.openprivacy.org/ • 23. Mutton, Paul and Jennifer Golbeck, "Visualization of Semantic Metadata and • Ontologies, " Proceedings of Information Visualization 2003, London, England, • July 2003. • 24. RDFWeb: FOAF: ‘the friend of a friend vocabulary’, http://rdfweb.org/foaf/ • 25. RDFWeb: Co-depiction Photo Meta Data: http://rdfweb.org/2002/01/photo/ • 26. Spertus, E, "ParaSite: Mining structural information on the Web," Proc. 6th • WWW Conf., 1997. • 27. Szalay, A. S. 2001, "Astronomical Data Analysis Software and Systems X," in • ASP Conf. Ser., Vol. 238, eds. F. R. Harnden, Jr., F. A. Primini, & H. E. Payne • (San Francisco: ASP), 3. • 28. The Trust Ontology: http://www.mindswap.org/~golbeck/web/trust.daml • 29. Watts, D. and S. H. Strogatz. "Collective Dynamics of Small-World' Networks", • Nature 393:440-442 (1998) Sharif University of Technology Computer Department

27. Questions? Sharif University of Technology Computer Department