Architecture of theWorld WideWeb - Social Semantics: The Search for Meaning on the Web - page 46

Web trades link consistency guarantees for global scalability” (2002). So, broken
links and 404 Not Found status codes are purposeful features , not defects, of
the Web.

2.3.3 Principle of Self-description

One of the goals of the Web is for resources to be 'self-describing,' currently defined
as “individual documents become self-describing, in the sense that only widely
available information is necessary for understanding them” (Mendelsohn 2006).
While it is unclear what “widely-available” means, one way for information to be
widely-available is for it to be linked to from the Web representation itself. The
Principle of Self Description states that the information an agent needs in order
to have an interpretation of a Web Representation (resource) should be accessible
from the Web representation itself (URI) .
How many and what sort of links are necessary to adequately describe a resource?
A resource is successfully described if an interpretation of a sense is possible. Any
representation can have links to other resources which in turn can determine valid
interpretations for the original resource. This process of following whatever data
is linked in order to determine the interpretation of a URI is informally called
'following your nose' in Web architecture.
The Follow-Your-Nose algorithm states that if a user-agent encounters a repre-
sentation in a language that the user-agent cannot interpret, the user-agent should,
in order:

1. Dispose of Fragment Identifiers: As mandated (Berners-Lee et al. 2005),
user-agents can dispose of the fragment identifier in order to retrieve whatever
Web representations are available from the racine (the URI without fragment
identifier). For example, in HTML the fragment identifier of the URI is stripped
off when retrieving the webpage, and then when the browser retrieves a Web
representation, the fragment identifier can be used to locate a particular place
within the Web representation.
2. Inspect the Media Type: The media type of a Web representation provides a
normative declaration of how to interpret a Web representation. Since the number
of IETF media-types is finite and controlled by the IETF, a user-agent should be
able to interpret these media types. 14
3. Follow any Namespace Declarations: Many Web representations use a generic
format like XML to in turn specify a customized dialect. In this case, a language
or dialect is itself given a URI, called a namespace URI , a URI that identifies that
particular dialect . A namespace URI then in turn allows access to a namespace

14 The finite list is available at http://www.iana.org/assignments/media-types/ , and a mapping from
media types to URIs has been proposed at http://www.w3.org/2001/tag/2002/01-uriMediaType-9 .