On Resources, Information Resources and Documents

It is crucial to understand the relationships of how the Web tries to describe and represent, or model, the world it is a part of. This is especially critical, since the most recent step of the evolution of the Web as Semantic Web is being realized. It shall enable machines to try to “understand” knowledge representations. Therefore, the terms Resource, Information Resource and Document are defined and their relationships are described in this article (see the illustrating graphic below). They are especially important to understand knowledge representation as it is also propagated in the Web.
This note is an attempt to explain my point of view regarding distinguishing the terms . At the beginning, I would like to define these terms. On this grounding, I would like to explain their relationships afterwards.

Resources

A Resource can be anything, i.e., a Resource is an entity regarding metaphysical terms. So, a Resource can be both a universal and a particular. It can be abstract or concrete – a concept, a categorization or a type (, a kind, a property or a relation), or an exemplification, an instantiation or a concrete object (cf. [1]). Following this differentiation, abstract objects are exemplified by concrete objects, and instantiations are categorized by concepts. Additionally, some particulars can be treated as universals for other particulars and recursively.
In terms of the Semantic Web, a Resource can be typed as rdfs:Resource, i.e., it can be characterized by a Description that is related to this CURIE (Compact URI). One can use a URI to denote a Resource, e.g., to name it in the Web. It is utilized for identification purpose in that context. So, a URI can be applied to identify a Resource¹ in some way (cf. the synonyms of identify in [11]). One can call such a URI Resource URI.

Information Resources

Before I continue the definition of the issued terms of this note, I have to define a further term, which will be used by other concept specifications. This term is called Semantic Graph. It is a formal knowledge representation (a specific kind of Description) for concepts, particulars and their relations. On the one side, an ontology can be represented by a Semantic Graph, e.g., a Semantic Web ontology. On the other side, instantiations of universals of ontologies can be represented by a Semantic Graph, e.g., a Named Graph, too.
An Information Resource is a piece of information that describes (or represents) a Resource. It includes this amount of information that is necessary to describe a Resource in a certain kind. Hence, the Subject² of an Information Resource is this characterized Resource. In general, an (abstract) Information Resource is usually realized by a Semantic Graph, which consists of a (concrete) Description of a Resource. This is necessary to make an Information Resource easily machine-processable. Of course, plain text is (more or less) easily processable by a human being, but not by a machine.
An Information Resource represents a Resource in some way. Nevertheless, it can not really represent a Resource completely. This is simply grounded in the nature of things that we can not really define what a complete Description shall be about. There exist at least always subjective Descriptions, which are unforeseeable.
In Semantic Web terms a Semantic Graph at least consists of one triple of the form: subject, predicate, object, however, often by a couple of triples. Such a triple is also called statement. To be more concrete, an Information Resource can be

a Minimum Spanning Graph [2], or
an RDF Molecule [3] (cf. [4]), or
a URI Declaration, or
a Concise Bounded Description, or
a resource-description, or
an associated description.

It depends on the definition one specifies for “the amount of information that is necessary to describe a Resource in a certain kind” (see above). An example of such a definition is the following set:

statements that have the Resource URI as object + dereferencing the subject + predicate URI of that statement for human-readable titles or names
+ statements that have the Resource URI as subject + dereferencing the object + predicate URI of that statement for human-readable titles or names
+ (optionally) include all information of the statements of these object Resources recursively, if they are part of this Semantic Graph, i.e., include at least all blank node objects³

A Resource can have multiple Information Resources. Each of them is embodied as a Representation⁴ that is delivered by dereferencing a Resource URI (cf. first definition of embody in [12]). That is why, a Resource can have multiple Resource URIs. Every Resource URI belongs to one Information Resource⁵. Due to that reason, a Resource URI can be an alias for another Resource URI, because there can exist different Descriptions (Information Resources) of one and the same Resource.

Documents

A Document can be used to deliver an Information Resource to an Information Consumer (as a specific Information Service⁶). Therefore, it should include a Semantic Graph, which consists at least of a Semantic Graph of that Information Resource. In other words, a Document contains at least one Information Resource. However, a Semantic Graph of a Document can include further information – in Semantic Web terms further (RDF) statements. Thus, a Document is that envelope that is sometimes be needed to represent or deliver an Information Resource to an Information Consumer.
It is a concrete thing, a particular. A Document is a specific Representation, e.g., a computer document. Such a sequence of bits at least has a content type, for instance application/xhtml+rdfa, and a name, such as a Document URI, for example, http://example.com/test.html.
Albeit, some Information Consumers do not need this Document envelope. They can process, e.g., a Semantic Graph of an Information Resource without further information. Such Information Consumers simply get a serialized version of the Information Resource of a requested Resource URI in an appropriated Representation format, e.g., Notation 3.

Resource Description Level Representations

An overall expectation of an Information Consumer is that an Information Provider (as a specific Information Service⁶) should deliver an Information Resource as response to a requested Resource. To transfer this statement to the Semantic Web one can say:

“The Web of Things is built on top of a Web of (Realizations of) Information Resources (that are delivered by Information Services)” (cf. [5]).

Following this definition, an Information Resource can have multiple Representations that at least embody a Semantic Graph of that Information Resource. A Resource URI and a Document URI can be the same, if a Resource is a Document. However, the relationships of the resource description levels can be different in this context. Dereferencing that Resource URI can deliver the Information Resource that describes this Document

in a computer document that has the Document URI as a name (a self description).
in another computer document that has another Document URI as a name.
in another Representation format.

All in all, at any time we should not directly infer knowledge about a Resource from Resource URIs, but always from Semantic Graphs (cf. Section 2.5 in [14]). For this interpretation task we ought to utilize especially those concrete Descriptions of the related Information Resources. These abstract Descriptions are addressed by the Resource URIs of a Resource. In that context, Resource URIs are simply present for dereferencing their Information Resources. Good URI design may sometimes help an Information Consumer to find a requested Information Resource. However, these interpretations are always expectations and are not fundamentally definitive.
The graphic below illustrates the relationships of the described terms.

Finally, one can say that everything can have an Information Resource (at least one).

Is a “real” thing, e.g., a car or a tree, itself “only” a Representation and hence a self description (Information Resource) of it? – If this is the case, then we can also claim, by insisting in Representationalism, that everything (every Resource) is an Information Resource.
For example, a human brain dereferences a Resource to an Information Resource in this context. In other words, one can simply identify a Resource by its Description (, which consists, e.g., of a (maybe quite difficult to imagine) Semantic Graph that connects and references other Resources). For me, this is an activity we are doing all the time. Thereby, some Resources occur by our understanding as “real” things.
Due to that conclusion, I guess, that I follow, from my understanding of Aristotle’s view, more then ever since before the Aristotelian realism. I can now conclude that every human being is a particular and a universal in its own, because one is both a self description and a self exemplification. Furthermore, one can insist the birth of a human being as an exemplification of/an attempt to exemplify the universals of the parents⁷. Ambiguousness exist, because we simply have at a specific temporal and spatial point only a partial understanding of things we perceive⁸. In that context, we cannot further reason to get the essence of a thing. So, it is a natural sensation, which is not bad at all. It is even often very helpful. However, it does not prevent us to reach a level of unambiguousness, which is not always necessary (cf. [15]).

(Tim Berners-Lee has a definition of abstract document (cf. [6]), which he calls Information Resource. In that characterisation, a Document is equal to an Information Resource. However, I would only treat a concrete document with a concrete content type etc. as Document, because otherwise the use of document is often misleading. Albeit, I more or less agree with his Information Resource definition. However, I would exclude the a priori equivalence of Document and Information Resource from this definition (, if he really stated this equivalence statement – [update]yes⁹[/update]).

please keep in mind:
“… the separation of layers … is fundamental” (from [7] page 4)

[update2]

Here is a first attempt of my abbreviated definition of Information Resource:

“An Information Resource is a Resource which can convey or describe (essential) characteristics of a Resource in some way, e.g., in a Semantic Graph. This Description can, for example, be realized (or embodied) as a concrete message, e.g., a serialisation (Representation) of a Semantic Graph in Notation 3 syntax. The Resource can also be the Information Resource itself, in which case it is referred to as a self description.”

It is aligned to the definition of ‘information resource’ that is given by the TAG (Technical Architecture Group) and Harry Halpin’s reflection of this TAG definition (see [13], p. 101). So, an Information Resource itself is more like a piece of abstract information [13], rather than a concrete Document. Albeit, for simplification purpose the term ‘information resource’ is often used to refer to both abstract information and particular Realizations (cf. [13]).
[/update2])

¹) Even though, other Resources can contribute to this identification process, e.g., a Description that has as (a) topic the intended Resource.

²) Here Subject should not be considered in terms of the subject position of a triple in an RDF statement.

³) Resource URIs that are part of other Semantic Graphs are dereferenced only once for human-readable label retrieval.

⁴) Representation can be defined as “data that encodes information about resource state” [14], whereby “information about resource state” can be seen as Description.

⁵) Equivalence of Information Resources and follow-up inferences that two Resource URIs denote the same Resource is another big issue, which can not be clarified in this scope. For example, a Resource URI can be related to an Information Resource that do not describe the essence of a Resource. Hence, two such Information Resources are ambiguous (cf. definition of substances in [1] and URI Declarations).

⁶) See [10] for an explanation of Information Service.

⁷) Please do not understand this claim in a wrong context. It should be seen from a full philosophical point of view.

⁸) Perception is/is a kind of/can be seen as dereferenciation or vice versa.

⁹) “I will use the words document and resource interchangeably … and sometimes may slip into using ‘object’.” (from [8]). Furthermore, the ambiguous use of the term Document in [9] is arguable.

[1] Loux, Michael J.; “Metaphysics – A contemporary introduction” (Second edition); Routledge; 2002
[2] Zhuge, Hai et al.; “Basic operations, completeness and dynamicity of cyber physical socio semantic link network CPSocio-SLN”; Wiley; 2010
[3] Ding, Li et al.; Tracking RDF Graph Provenance using RDF Molecules; UMBC Tech Report TR-CS-05-06; 2005
[4] Berners-Lee, Tim; “Linked Data”; w3.org; 2006 – 2009
[5] Berners-Lee, Tim; “Levels of Abstraction: Net, Web, Graph”; w3.org; 2007 – 2010
[6] Beners-Lee, Tim; “Ontology for relating Generic and specific Information Resources”; w3.org; 2009
[7] Berners-Lee, Tim; “Long Live the Web: A Call for Continued Open Standards and Neutrality”; scientificamerican.com; November 2010
[8] Berners-Lee, Tim; “Metadata Architecture”; w3.org; 1997 – 2009
[9] Berners-Lee, Tim; “Meaning”; w3.org; 1999 – 2009
[10] Ferris, Bob; “What is an Information Service?”; infoserviceonto.wordpress.com; 2010
[11] several; “identify”; dict.cc; 2010
[12] several; “embody”; thesaurus.com; 2010
[13] Halpin, Harry; “Sense and Reference on the Web”; ibiblio.org; 2009
[14] Jacobs, Ian et al.; “Architecture of the World Wide Web, Volume One”; w3.org; 2004
[15] Hayes, Patrick J. et al; “In Defense of Ambiguity”; ibiblio.org; 2008