5
ShareNet Integrating Trust and Privacy policy Li Ding

ShareNet Integrating Trust and Privacy policy Li Ding

Embed Size (px)

Citation preview

Page 1: ShareNet Integrating Trust and Privacy policy Li Ding

ShareNetIntegrating Trust and Privacy policy

Li Ding

Page 2: ShareNet Integrating Trust and Privacy policy Li Ding

The Research Road MapRepresentation• Web entity

• Individual: person, website, robot• Community: social network, fiends

• Complex knowledge• relation: trust, proof, provenance• rule: policy

• Others: logging, web credibilityComputation• Distributed co-learning • Network/graph analysis• Distributed logical inference

Technology• Web service: WSDL, OWL-S, SOAP• Knowledge creation: auto translation, XSLT• Knowledge representation: P3P, RSS, FOAF• User interface: XSLT

Page 3: ShareNet Integrating Trust and Privacy policy Li Ding

Roadmap

• Test --Privacy Policy Sharing– Framework – Context details

• Ontology– Address– FOAF-Lite– WebOfBelief

• Association• Assertion• AssertionProb

– Website• Privacy policy• Shopping rating

– Model/ Rule• weightedModel

• Agents– Web service

• Pass OWL content via SOAP as (attachment ) (no in SOAP body)

• Create multiple instance of one web service

– How to express query• Jena query• Tripple

– Roles• Person• P3P converter• Google • Amazon• reputation

Page 4: ShareNet Integrating Trust and Privacy policy Li Ding

P2P user network

Web Information sources

Robots

Testbed Framework

Facilitator

PersonalWS

GoogleRWSReputationWS

EpinionRWSProxy

PersonalWS

PersonalWS

Page 5: ShareNet Integrating Trust and Privacy policy Li Ding

Privacy Policy Sharing

Context• M=50 users and N=100 websites• “know” relations is

– Randomly initialized: each user randomly know u users, and u follows (normal, zipf) distribution.

– Randomly connected groups: users in the same group knows one another, then users are randomly connected

• “knowledge about website” – Range is “yes, no” – if the website has

privacy policy– May not knowing the website– Rating

• “trust” relation is– Dynamically learned from experience– Dynamically inferred from network

Scenarios• Proxy/TestAgent ask user A about

website X via facilitator• Testing Agent generate initial

knowledge distribution and send them to each personal agent

• Personal agent outsource knowledge/inquire rating

• Personal agents use their models (utility function) to make decision

• Personal agents evolve trust knowledge– QueryifWebsite: with α probability use

own knowledge, otherwise use consensus

– InitKnowledge: “know”, “website rating”, “trust evolution choice”

– Register