Conclusion and Open Issues

The paper motivates the need to incorporate semantic metadata in file stores. We identify the basic types of semantic information required by applications and end-users and propose a generic data model to capture and represent file semantics. The model provides the basis for a framework of tools and APIs for generating and using semantic metadata. There is a large number of research problems that need to be addressed to realize a semantic-aware file store. We enumerate some of them below.

• The basic semantic relations sketched in section 2.2 are yet to be evaluated and finalized through the use of real applications.

• Investigate the design of semantic-aware deep-archival systems. In particular, what kind of semantic information can be used for improved data clustering and compression techniques. Also, how to maintain rich semantics for multiple versions of files; inheritance of semantic relations and their representation and use.

• Use semantic metadata for intelligent data placement in distributed storage systems. The goal is to satisfy the QoS requirements of end-users or applications with low infrastructure cost.

• Design and implement a basic set of tools and APIs for using the semantic information captured in such systems. These tools should be extensible and customizable. What these tools will be and how they will interact with each other is an open issue.

• Devise a simple declarative query language that can be used to specify constraints on both structured and unstructured data components.

• Investigate how the proposed data model and framework can be implemented in a distributed file system efficiently. One hard question is how to store RDF relations using a lightweight DB.