CSAIL logo MIT logo

Secure Federation Systems for Semantic Web Sources

Overview

The Semantic Web is a powerful distributed model for publishing, utilizing, and extending information. It provides several enabling technologies and protocols such Resource Description Framework (RDF), RDF Schema (RDF-S), Web Ontology Language (OWL), and the SPARQL Query Language. These technologies as well as the current focus on Linked Open Data have led to several online Semantic Web data sources whose contents can be ``mashed-up'' (i.e. queried and integrated on-the-fly) to provide rich Web applications. Our challenge is to create open information environments that allow data from different sources to be seamlessly integrated while preserving appropriate security and privacy policies.

There are several hurdles that need to be addressed before dynamic mashups can be developed. A user or application developer needs to know exactly what data is present in each source and needs to formulate queries carefully in order to get a reply. When the content of a source changes, mashups/users cannot take advantage of it until the application developer becomes aware of this change and modifies the application/query accordingly. Performing open ended queries that span multiple sources requires the user to decompose the query according to the diff sources, send the queries to the appropriate sources, and then compose the result. One possible mechanism that meets these challenges is the federation paradigm from database systems. Federated database systems enables users and applications to issue a single query to the federation engine, which then converts it into multiple queries against distributed data sources, and returns the merged result of those queries.

Though the federation engine provides transparent access to multiple data source, the lack of a shared model for security and privacy requirements impedes this transparency as the federation engine is unable to process the different requirements of each data source and obtain appropriate credentials from the requester. This causes most federations to require prior setup and negotiation of policy and prevents the dynamic integration of data from sources that use different policy languages. We do not believe that users should be forced to conform the description of their policy relationships to a single standard policy language. Instead there should be a way of encompassing different policy languages and supporting heterogeneous policy systems. This project hopes to leverage the power of the Semantic Web to provide a common policy interlingua that will help preserve maximum expressiveness for policy communities by allowing users to define policies in their own languages but still enable inter-domain communication and data sharing via policy translation, reasoning, and enforcement.

This project is aimed at developing protocols and technologies to support secure federation systems for the Semantic Web that will enable large-scale dynamic aggregation across heterogeneous sources that use different policy languages.

Students

We are interested in working with students at all levels, UROP, MEng or PhD. Students interested in this research should come talk to us.

References
maintained by Lalana Kagal
 $Revision: 31051 $
$Date: 2011-06-28 14:50:00 -0400 (Tue, 28 Jun 2011) $