A Reflection on the Structure and Process of the Web of Data

Marko A. Rodriguez is director’s fellow at the Center for Nonlinear Studies, T-7, Los Alamos National Laboratory, Los Alamos, New Mexico 87545; Markolanl.gov T he web community has introduced a set of standards and technologies for representing, querying and manipulating a globally distributed data structure known as the Web of Data. The proponents of the Web of Data envision much of the world’s data being interrelated and openly accessible for use by the general public. This vision is analogous in many ways to the web of documents of common knowledge, but instead of making documents and media openly accessible, the focus is on making data openly accessible. Providing data for public use has stimulated interest in a movement dubbed Open Data [1]. Open Data is analogous in many ways to the open source movement. However, instead of focusing on software, Open Data is focused on the legal and licensing issues around publicly exposed data. Together, various technological and legal tools are laying the groundwork for the future of global-scale data management on the web. As of today, in its early form, the Web of Data hosts a variety of data sets that include encyclopedic facts, drug and protein data, metadata on music, books and scholarly articles, social network representations, geospatial information and many other types of information. The size and diversity of the Web of Data is a demonstration of the flexibility of the underlying standards and the overall feasibility of the project as a whole. The purpose of this article is to provide a review of the technological underpinnings of theWeb of Data as well as some of the hurdles that need to be overcome if theWeb of Data is to emerge as the de factomedium for data representation, distribution and ultimately, processing. Technically, on the Web of Data, Uniform Resource Identifiers (URI) are used to identify resources [2]. For example, depending on what is being modeled, a URI can denote a city, a protein, a music album, a scholarly article or a person. In fact, in general any thing can be assigned a URI. An example URI is . This URI denotes the author of this article,Marko. The URI has information pertaining to the what (marko), where (www.lanl.gov) and how (http) of a resource. The URI is more general than the commonly used URL, as URIs are not required to resolve to retrievable digital objects such as documents and media. Instead, URIs can denote abstract concepts such as the person Marko, the class of dogs or the notion of friendship. Finally, the space of all URIs is an inherently distributed and theoretically infinite space. This attribute makes the URI space fit to represent massive amounts of data distributed worldwide. A convenient consequence of this space is that the Web of Data can emerge atop it. However, while URIs can denote things, they cannot denote how things relate to each other. Relating URIs is necessary in order to give greater meaning and context to each datum. Moreover, relating URIs is necessary to create the web aspect of the Web of Data. The Resource Description Framework (RDF) is a standardized data model for linking URIs in order to create a network/graph of space of all URIs [3]. RDF also supports the linking of URIs to primitive literals such as strings, integers or floating point values. An example RDF statement to denote the fact that “Marko knows Fluffy” is . In order to make long URIs more readable, namespace prefixes are generally used. With namespace prefixes, the previous statement can be represented as lanl:marko, foaf:knows, lanl:fluffy. All RDF statements have this three Feature