A Semantic Web Middleware for Virtual Data Integration on the Web

A Semantic Web Middleware for Virtual Data Integration on the Web
A Semantic Web Middleware for Virtual Data Integration on the Web
Bibliographical Metadata
Subject:	Querying Distributed RDF Data Sources
Keywords:	not available.
Year:	2008
Authors:	Andreas Langegger, Wolfram Wöß, Martin Blochl
Venue	ESWC
Content Metadata
Problem:	SPARQL Query Federation
Approach:	Querying Distributed RDF Data Sources
Implementation:	SemWIQ
Evaluation:	Sample Queries Evaluation

Abstract

In this contribution a system is presented, which provides access to distributed data sources using Semantic Web technology. While it was primarily designed for data sharing and scientific collaboration, it is regarded as a base technology useful for many other Semantic Web applications. The proposed system allows to retrieve data using SPARQL queries, data sources can register and abandon freely, and all RDF Schema or OWL vocabularies can be used to describe their data, as long as they are accessible on the Web. Data heterogeneity is addressed by RDF-wrappers like D2R-Server placed on top of local information systems. A query does not directly refer to actual endpoints, instead it contains graph patterns adhering to a virtual data set. A mediator finally pulls and joins RDF data from different endpoints providing a transparent on-the-fly view to the end-user. The SPARQL protocol has been defined to enable systematic data access to remote endpoints. However, remote SPARQL queries require the explicit notion of endpoint URIs. The presented system allows users to execute queries without the need to specify target endpoints. Additionally, it is possible to execute join and union operations across different remote endpoints. The optimization of such distributed operations is a key factor concerning the performance of the overall system. Therefore, proven concepts from database research can be applied.

Conclusion

In this contribution a mediator-based system for virtual data integration based on SemanticWeb technology has been presented. The system is primarily developed for sharing scientific data, but because of its generic architecture, it is supposed to be used for many other Semantic Web applications. In this paper query federation based on SPARQL and Jena/ARQ has been demonstrated in detail and several concepts for query optimization which is currently on the agenda have been discussed. Additional contributions can be expected after the implementation of additional features mentioned before.

Future work

Other future work will be the support for DESCRIBE-queries and IRIs as subjects. In future, the mediator should also use an OWL-DL reasoner to infer additional types for subject nodes specified in the query pattern. Currently, types have to be explicitly specified for each BGP (more precisely for the first occurrence: the algorithm caches already known types). OWL-DL constraints like for example a qualified cardinality restriction on obs:byObserver with owl:allValuesFrom obs:Observer would allow the mediator to deduce types of other nodes in the query pattern.

Approach

Positive Aspects: {{{PositiveAspects}}}

Negative Aspects: {{{NegativeAspects}}}

Limitations: {{{Limitations}}}

Challenges: {{{Challenges}}}

Proposes Algorithm: {{{ProposesAlgorithm}}}

Methodology: {{{Methodology}}}

Requirements: {{{Requirements}}}

Limitations: {{{Limitations}}}

Implementations

Download-page: {{{Download-page}}}

Access API: {{{API}}}

Information Representation: {{{InfoRepresentation}}}

Data Catalogue: {{{Catalogue}}}

Runs on OS: {{{OS}}}

Property "RunsOn OS" (as page type) with input value "{{{OS}}}" contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.

Vendor: {{{vendor}}}

Uses Framework: Jena

Has Documentation URL: http://semwiq.faw.uni-linz.ac.at/core/2007-10-24/ catalog.owl.

Programming Language: Java

Version: {{{Version}}}

Platform: {{{Platform}}}

Toolbox: {{{Toolbox}}}

GUI: No

Research Problem

Subproblem of: {{{Subproblem}}}

Property "Has Subproblem" (as page type) with input value "{{{Subproblem}}}" contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.

RelatedProblem: {{{RelatedProblem}}}

Property "Has relatedProblem" (as page type) with input value "{{{RelatedProblem}}}" contains invalid characters or is incomplete and therefore can cause unexpected results during a query or annotation process.

Motivation: {{{Motivation}}}

Evaluation

Experiment Setup: The tests were performed with the following setup: the mediator (and also the test client) where running on a 2.16 GHz Intel Core 2 Duo with 2 GB memory and a 2 MBit link to the remote endpoints. All endpoints were simulated on the same physical host running two AMD Opteron CPUs at 1.6 GHz and 2 GB memory.

Evaluation Method : {{{EvaluationMethod}}}

Hypothesis: {{{Hypothesis}}}

Description: {{{Description}}}

Dimensions: Performance

Benchmark used: Kanzelh¨ohe Solar Observatory (KSO)

Results: Because ARQ is using a pipelining concept the response time is very good, even when data has to be retrieved from a remote data source.