A Probabilistic Approach to Information Control

Abstract

Information aggregation is identified as one of the key threats of the emerging Semantic Web. Although the new web has experienced slow deployment, the advent of Cloud Computing is assumed to boost the rate of deployment by providing means to conduct resource intensive analysis, tracing and querying of semantically annotated and linked data with minimal investment cost for suppliers. However, the threat of information aggregation poses a severe threat to privacy protection, business-critical information and some aspects of national security. The lack of security mechanisms to restrict access to and usage of information after it has been distributed, further contributes to the risk of aggregation since redistribution cannot be controlled. In this paper we propose a probabilistic approach to information control based on trust management systems. Our solution provides the user with a view of the amount of information that any given entity probably has received through redistribution (from others), in order to determine the level of aggregation the entity can perform. We define a middleware architecture and provide an implementation in a simulation environment. Initial results from experiments demonstrate the accuracy of the model and that there may be significant benefits from the approach in several application areas.