Linked Data Spaces & Data Portability

Abstract

In the year 2007, the size of the Linked Data injected into the Web grew to several billion RDF triples, served by a network of interlinked data sources that cover domains such as general knowledge, geographic information, people, companies, online communities, films, music, books and scientific publications. Unfortunately, the growth rate of User Generated content from a variety of Web based unstructured and semi-structured data-silos continues to exceed that of structured Linked Data. Thus, we have a pressing need for technology, capable of bridging this broadening divide via transparent generation of Linked Data from existing data-silos on the Web. Our Linked Data technology demonstration explores the use of the OpenLink Data Spaces platform as a solution to this problem.

Categories and Subject Descriptors

H.3.2 [Information Storage]

H.3.3 [Information Search & Retrieval]

General Terms

1. INTRODUCTION

User generated content is growing at an exponential rate behind

corporate firewalls and across the Internet in general. The use of

Web technologies has been the prime accelerator of the

aforementioned growth due to the pervasiveness of Web based distributed collaborative applications. Examples include: Social

Networking, Weblogs, Wikis, Shared Bookmark Managers, Photo

Sharing, Polls Management, Calendars, Discussion Forums, File

Sharing, and Feed Aggregation, to name a few.

The exponential growth of user-generated content has resulted in

the growth of silos comprised of unstructured and/or semistructured content. Unfortunately, these silos have accelerated, rather than decelerated, the imminence of an “information overload” quagmire. • •

RDF based structured data

Standardized data serialization formats

HTTP based Unique Identifiers for all Data Items (web

resources and abstract & concrete things)

HTTP based Data Set containers (Data Spaces)

Data Servers that provide data management and data

access services for one or more Data Spaces

Key infrastructure oriented shared ontologies

Query Language for interacting with structured data

We identify the items above, collectively, as critical components

of Linked Data Spaces: points of presence on the Web that expose structured data via HTTP based URIs.

2. Issues & Challenges

2.1 Data Portability

It’s no secret that data wants to be free of the tyranny of

application logic confinement. In recent times, the realization that meshing Identity and Data ownership on the Web are critical requirements of this pursuit of freedom, has resulted in the emergence of a movement for Data Portability as yet another enclave within the broader Open Data movement.

Data portability addresses to key issues: data mobility and data referencing. Today, data mobility though the use of standard data formats for moving data across silos (import and export style) have emerged as the focal point of attention with regards to addressing the proliferation of data silos on the Web. Examples include: RSS 1.0, RSS 2.0, Atom, OPML, FOAF, SIOC, and others. Unfortunately, the ability to reference and de-reference data across data-silos is yet to catch the attention of those pursuing data portability.

2.2 RDFization Middleware

The traditional resistance to RDF adoption, which is critical to

Linked Data comprehension and production, comes from the

grounding of the RDF Data Model in Graph Theory and the unwillingness of most Web Application developers to interact with data formally. This reality has lead to a genre of middleware tools collectively known as RDFizers, that generate RDF on the fly.

With regards, to Linked Data, generating RDF on-the-fly is only part of the equation; the generated RDF must retain the core principles of linked data by providing URIs for physical web accessible resources, concrete entities, and abstract things. Of course, this process must include intelligent production of instance data associated with relevant shared schemas or ontologies.

2.3 Data Junction Boxes in the Clouds

It is our belief that the Linked Data Web will be more distributed

than centralized in architecture. We envisage a Linked Data Web comprised of hubs that range is size from large (e.g. DBpedia,

Geonames, Zitgist etc.), medium sized group (e.g. RDFized

Weblogs, Wikis, Bulletin Boards etc.), and smaller personal hubs

enabled by operating system virtualization technologies like

Amazon EC2. The medium and smaller hubs are best described as data junction boxes because they act as conduits between existing systems and Linked Data aware User Agents.

This demonstration will demonstrate a Data Space initialization process for end-users that covers:

• • • •

Domain Name Registration (e.g. .Name acquisition)

DNS configuration

Bonding with existing Web 2.0 platforms Facebook,

phpBB3, MediaWiki, Wordpress, Drupal, Del.icio.us,

Flickr, and Bugzilla

Production of a dereferencable URIs that exposed the resulting Data Graph

•

Interaction with the resulting data graph via a number of

Linked Data aware User Agents

4. Links • • • http://en.wikipedia.org/wiki/OpenLink_Dat a_Spaces - OpenLink Data Spaces http://en.wikipedia.org/wiki/Virtuoso_Univ ersal_Server - Virtuoso http://myopenlink.net/ods/index.html - Live OpenLink Data Spaces Demonstration

References

3. Identity & Data Meshing via Linked Data Spaces