Stephen Hearn: November 2010 Archives

Linked Data/Semantic Web

user-pic
Vote 0 Votes

The term "linked data" refers to a particular model of encoding relationships between entities, concepts, and things. The model consists of a simple "triple" syntax. Linked data statements contain a subject, a relationship, and an object. These three terms are expressed in standard encoding, usually RDF (Resource Description Framework). For example:

Thomas Pynchon [subject] is author of [relationship] Gravity's Rainbow [object]

What really makes linked data work, though, is the substitution of URIs for each of the three terms. The URIs function as consistent controlled identifiers for the entity, concept, or relationship they represent, eliminating the variability that pervades natural language expressions and enabling data statements to be confidently and specifically linked.

Linked data can be used both within a data environment and across data environments. Within an environment it makes possible the automatic generation of pages displaying relationships among entities and concepts. When one data environment makes use of URIs from another environment, additional kinds of functionality and integration can be achieved. In this way, URIs can expand in use beyond their home environments to many other data sets, and become the basis of data assembly across larger and larger pools of data. Therefore, linked data becomes part of the Semantic Web.

Links:
Examples of sites built on linked data:
DBpedia is a compilation of linked data statements derived from Wikipedia. It's designed primarily for machine-to-machine interactions, but knowing the basic URI structure lets you call up sets of data harvested around a given topic, e.g., the University of Minnesota above. Not as pretty as the first two sites, but a clearer representation of the underlying linked data.