120 likes | 239 Views
Using Trust and Provenance for Content Filtering on the Semantic Web. By Jen Golbeck & Aaron Mannes Maryland Information Network Dynamic Lab University of Maryland, College Park. What are social networks. Connections between people Can be Explicit (people say who they know)
E N D
Using Trust and Provenance for Content Filtering on the Semantic Web By Jen Golbeck & Aaron Mannes Maryland Information Network Dynamic Lab University of Maryland, College Park
What are social networks • Connections between people • Can be • Explicit (people say who they know) • Derived (e.g. from email archives) • Simulated
Web-Based Social Networks (WBSNs) • Websites and interfaces that let people maintain browsable lists of friends • Last count • 142 social networking websites • Over 200,000,000 accounts • Full list at http://trust.mindswap.org • Over 10,000,000 accounts are represented in FOAF, an OWL ontology
Trust in WBSNs • People annotate their relationships with information about how much they trust their friends • Trust can be binary (trust or don’t trust) or on some scale • This work uses a 1-10 scale where 1 is low trust and 10 is high trust • At least 8 social networks have some mechanism for expressing trust
Inferring Trust The Goal: Select two individuals - the source (node A) and sink (node C) - and recommend to the source how much to trust the sink. tAC A B C tAB tBC
Trust Algorithm • If the source does not know the sink, the source asks all of its friends how much to trust the sink, and computes a trust value by a weighted average • Neighbors repeat the process if they do not have a direct rating for the sink
Film Trust • Working example of this can be found at - FilmTrust available at http://trust.mindswap.org/FilmTrust • A movie recommendation site backed by a social network that uses trust values to generate predictive recommendations and sort reviews
Applications of Trust • With direct knowledge or a recommendation about how much to trust people, this value can be used as a filter in many applications • Since social networks are so prominent on the web, it is a public, accessible data source for determining the quality of annotations and information
Trust Networks & Intelligence • Intelligence agencies no longer face hierarchies, now they face networks • Several major intelligence failures due to lack of information-sharing or adequately questioning dominant assumptions • Sheer size of intelligence communities are often a barrier to information sharing • Trust networks could help intelligence agencies connect the dots
Use Case Scenarios • Help individual analyst sort through mass of material by identifying reliable sources • Trust ratings would allow analysts to check veracity of information by seeing how sources are rated by other trusted analysts • Importance of outliers for red-teaming - a team comes to strong conclusions on an issue: policy-makers could use trust ratings to check with dissenters
Uses for Meta-Data • Analyzing patterns of trust ratings could help break organizational barriers • While outliers are useful on a case by case basis they could also indicate an organizational dysfunction • A pattern of low trust ratings between units could indicate a conflict or lack of understanding • Alternately a pattern of particularly high ratings could indicate group think
References • Papers and software available at http://trust.mindswap.org • FilmTrust available at http://trust.mindswap.org/FilmTrust • Terrorism Analysis available at http://profilesinterror.mindswap.org/ • golbeck@cs.umd.edu • awmannes@comcast.net