OGSA-DAI-RDF

DAI-RDF Overview †

Introduction †

DAI-RDF is a set of RDF processing OGSA-DAI activities including a query processing activity using W3C SPARQL.

The feature of OGSA-DAI-RDF is this R&D project is synchronized with OGF DAIS-WG discussion of WS-DAI realization for RDF(DAIS for RDF). DAI-RDF aims to be one of reference implementations for the coming WS-DAI-RDF(S) standard.

The current supported RDF data repositories are Jena2, Sesame. Boca2 will be also supported. The software is preliminary released in 2006 and the first release with stable spec is released in 2007.05. New version for DAI3.0/3.1 is released in 2008/2009

What's New in V1.0 †

• Re-Designed Activity Interfaces
  • Support of Piped data and References for I/O data.
  • Introduction of product-independent driver interface
  • WS-DAI-RDF Querying demonstrative sample codes(3.0 version only)
• New GUI including Activity Workflow Editor(2.2 version only)

Architecture †

As shown in the following picture, DAI-RDF provides internal product-independent access layer which wraps product dependency. We provided several RDF repository drivers for OGSA-DAI-RDF. This interface also supports the http server interfaces which many existing RDF repository provides.

RDF Data Resouce †

In DAI-RDF, an RDF data resource in the context of OGSA-DAI is equal to a RDF Dataset of W3C SPARQL. An RDF Dataset is a set of RDF Graphs. An RDF Graph is a set of RDF triples which has Subject, Predicate and Object. Each graph is uniquely identified with URI. Thus, an RDF data resource holds a set of Graphs with URI.

To support RDF Dataset, DAI-RDF introduces new resource type "RDF" to the original OGSA-DAI resource types such as "XML" and "Relational". Thus, several heterogeneous data resources including RDF data resources can be handled with a single OGSA-DAI data service.

The RDF data resource can be implemented with different RDF data repository such as Jena or Sesame. One RDF data resource corresponds to one repository product. Thus, a OGSA-DAI data service which can holds multiple RDF data resource can handle heterogeneous RDF data repositories.

Most of RDF repositories provides several implementation types for holding RDF triples. For instance, Jena2 model can be on temporary memory or on persistent rdb. Thus, we introduced storage type where RDF resource is created.

Structure of RDF Processing Activities †

Basically, all DAI-RDF activities supports following input/output types.

• URI reference: The identifier to the data to be processed.
• Data Stream: Piped data stream itself.

For instance, sparqlQueryStatement Activity which performs SPARQL query processing has following input parameters for RDF data.

This means that users can make SPARQL query to the some stored data identified with URI and to the data which is sent from other activities.

An example of this structure is the reasoner with query as shown in the next figure.

Activity Framework with other OGSA-DAI provided activities. †

OGSA-DAI activity framework provides to construct a complex data processing by combining several data processing activities. OGSA-DAI-RDF activities can be combined with other OGSA-DAI activities so that it is easily possible to construct a distributed RDF data processing by combining OGSA-DAI data conversion and transfer activities. This perform document is a simple example which sends the result of SPARQL query to another site.

Current provided activities for RDF †

• ./Activity Interface/DataSetManagementActivity : The interface for RDF Dataset(=Set of RDF Graphs) This function provides to create,remove and list RDF Graphs within a DataSet(=RDF data resource)

• ./Activity Interface/GraphManagementActivity : The interfece for RDF statement. This activity provides to Insert and Delete RDF statements(Triples).

• ./Activity Interface/sparqlQueryStatementActivity : The interface for SPARQL query interface. Named graphs and Default graphs are supported.

• ./Activity Interface/OntologyReasonerActivity : This interface invokes a reasoner and create an infered graph. The output can be RDF data stream or graph URI which identifies the created graph.

• rdfsSchema Activity
• rdfsInstance Activity

These two activities provides Ontology access functions based on RDF(S) and OWL. These activities are provided by our preliminary release, however, this fuction will depends on the current WS-DAI-RDF Ontology Access specification and we stopped to provide this interface until some fixed direction is presented within DAIS-WG discussion.

Future Plans †

• Support of more RDF platform software
  • RINZA
• Support of More Activities
  • Seccurity
  • Versioning
• Distributed Processing
  • Extension of OGSA-DQP
• Support of WS-DAI-Ont Spec

DAI-RDF Application within AIST †

• S-MDS Semantic MDS resource monitoring and discovery
• Semantic SOA
• GeoGrid Metadata Registry