Models and framework for metadata generation and policy infrastructure

Enforcement of data sharing agreements requires that the security requirements of the agreement are expressed as enforceable policies, and these policies can be evaluated in a cross-domain scenario. This document focuses on policy infrastructure and metadata generation infrastructure to support these requirements. Organizations depend on data sharing for varying reasons. While they need to share data, they want to control how the shared data is used after it has been given to another organization. The requirements to control usage of data may originate from several requirements, like the need to protect sensitive data, protect commercial interests of the organization; provide accountability, or the need for timely dissemination of data. The problem of sharing data in cross-domain scenario presents a combination of several problems, like access control, rights management, trust management, privacy, etc. A key requirement for controlling the usage of data is to provide continuity of control. This problem is called usage control. In past, research has focussed mainly on individual problems mentioned above. There is a lack of architecture that allows flexibility to addresses these problems together. We have given an overview of architecture to address these needs in deliverable document D1.1. This document describes policy infrastructure that is needed to support the Consequence architecture. We discuss an enforceable policy language capable of expressing the requirements seen in the testbed scenarios. Note that in this document we always talk about enforceable policies even when we may not say so explicitly each time. We also identify components of the policy infrastructure and discuss how they should interact with other components of the Consequence architecture. Meta-data is an important ingredient of policy infrastructure. It can be used to describe the characteristics of data, like who owns the data, how the data was collected, what the data is about, etc. It can also be used to describe an organizations structure, users, and their attributes. In the policy infrastructure, metadata plays the role of glue that binds policies with the protected data and its users. To enable enforcement of policies across organizational boundaries, the organizations sharing data should agree on a common metadata vocabulary and associate metadata with protected data. Therefore, in this document we also discuss the requirements for metadata generation and propose a methodology to be used in Consequence. The rest of the document is organized as follows. In Section 2, we first discuss the requirements for policy infrastructure and metadata generation to support the goals of data sharing and usage control. We also identify requirements that should have higher priority in development of policy infrastructure. In Section 3, we present related research work and discuss their shortcomings. Finally, in Section 4, we propose a policy language, components of the policy infrastructure, and infrastructure for metadata generation. We also analyze features of the proposed models against the requirements identified in Section 2.