Distributed Semantic Metadata Repository

The Vision:

The IT infrastructure is quickly moving from stand-alone applications to a more integrated world in which several applications are linked together and cooperating to solve more general problems. In a changing economic climate, the agility of solutions becomes a key asset, requiring a move towards the dynamic composition of applications, web services and an adaptive management.

To make this possible, the first and probably most difficult problem to solve is to repurpose the data between legacy applications and services, allowing each of them to access, share and transform the data without modifying the original applications.

The widely deployed XML format has become the de-facto data interchange format. However it has a weak data model incapable of capturing semantics, relationships or constraints.

Our vision is to introduce a thin semantic modelling layer on top of existing XML applications to allow semantic interoperability. Our representation language is RDF (Resource Description Framework, a W3C open standard currently under development), offering great modelling possibilities because it is unstructured and is a complement to the semi-structured XML format. In addition, this metadata layer allows exploration of a new data access metaphor, probably leading to yet-to-be-discovered querying mechanisms to be used in future Semantic Web applications.

Problems Addressed:

Create a semantic metadata repository for static and dynamic metadata integrating existing data stores to:

• Model data and metadata; transform, repurpose, and store into databases
• Offer advanced reasoning on metadata
• Hide complexity of metadata from users; allow "semantic queries" mixing reasoning and querying
• Improve agility of systems
• Bridge gap between databases, dynamic content and dataflow

Our Contribution:

The main contributions from HP Labs in this domain is the design and implementation of Genesis, a distributed semantic metadata repository, which focuses on deployment and distribution issues. It achieves two main goals:

• Provide associative storage that supports evolving schemas for semi-structured data, enabling sophisticated search mechanisms above semi-structured metadata, and data repurposing
• Provide key mechanisms for collaboration between DB and groups of users in a distributed environment

In parallel to this research, HP Labs,

• realized the permanent storage of Jena2. Jena is the Java-based open source Semantic Web toolkit developed by HP Labs
• continues work on a benchmark for RDF.

Applications

The Adaptive Enterprise and Web Services all rely on a data repurposing layer allowing existing applications to communicate and interact. We are introducing ontological representations and metadata to enhance agility and favour a seamless management of heterogeneous systems.

We are exploring new ways for digital libraries to store, achieve and retrieve digital assets, in an environment where data will evolve over time. We participate to the SIMILE project where HP, W3C and MIT join forces to investigate into new areas of information management.

Contact:
Bernard Burg
email: bernard.burg@hp.com
phone: +1 (650) 857 2403