dimanche 20 avril 2008

semantic web, social web, web 2.0, web 3.0, ... web n.0

By nature, a Web application requires taking into account its semiotic dimension (as a meaningful system mobilizing signs of all types to build representations for humans), its pragmatic dimension (as a semiotic system with multiple usages which influence its interpretation) and its social dimension (as a virtual space of interaction). In our opinion, nothing in the semantic Web initiative is opposed to taking these dimensions into account quite to the contrary we might add. The official semantic web activity page says "The Semantic Web provides a common framework that allows data to be shared and reused across application, enterprise, and community boundaries" . The core motivating scenario of the semantic web initiative thus is assistance to collaboration.

Over the past eight years, semantic web researchers have proposed pivot languages to represent and exchange data, but they have prescribed or restricted neither the use of these languages nor the methods to generate the data they convey. Opposing ontology and folksonomy is like opposing a "cake" and a "baked cake" - they are not at the same level of abstraction. Folksonomies are defined by the way they are obtained (social tagging). Ontologies are defined by their content (a representation of a conceptualization), and not by the way they are obtained. Furthermore, nothing in the objectives or formalisms of the semantic web is opposed to taking into account the social dimension of the web.

Another important point that is often missed is the notion of domain of formalization. In a semantic web application, the domain to be formalized in RDF/S or OWL is not always the application domain since formalization is primarily specified by the task. Thus if the task is, for instance, assistance with the alignment of several medical terminologies, perhaps formalization will focus on linguistic primitives (e.g. term, synonymous with, hyponym of, acronym of, etc.), allowing the representation and comparison of the various terminologies. In other words, we do not inevitably find the notions of our domain of application in our RDFS schema. Building semantic web applications in the automotive industry doesn't necessarily mean we will find the concept of "car", "windshield" or "engine" in our ontology. In a Web 2.0 application this means that the use of semantic web frameworks does not imply that the domain of application will have to be formalized. As an example, if I want to allow social tagging for a music bank I don't necessarily have to formalize music categories in RDFS but I can choose to formalize the domain of social tagging and declare the notions of "tag", "tag cloud", etc.

Another dangerous misconception is to consider that ontology-based solutions are necessarily centralized and/or monolithic solutions with frozen schemas. Multiple schemas, namespaces and equivalence relations are examples of core mechanisms of semantic web formalisms that clearly show that the semantic web vision is that of a distributed, decentralized, integration system. Moreover, to represent and publish an ontology in RDFS or OWL does not imply that it is now set in stone. The semantic web perfectly acknowledges the existence of a life-cycle of ontologies (e.g., relations of equivalences between two ontologies in OWL, best practices to choose URIs to manage the evolution of a concept, etc.).

Finally, designing an application using semantic Web frameworks does not imply building a solution only with the tools of the semantic web. On the contrary, a proprietary application which manages an electronic calendar is an application of the semantic web if it simply makes it possible to export and to import its data in one of the languages of the semantic web. It doesn't need to have a rule engine, a Prolog virtual machine, some tableau algorithms or a projection operator implementing SPARQL. The only effort which is required from it is to do the only thing towards the interoperability which cannot be done by something else to make explicit its data structures and the conceptualization on which it is based. It is the old challenge of ontologies but with results at the scale of the Web.

from text I contributed in Michel Buffa, Fabien Gandon, Guillaume Ereteo, Peter Sander and Catherine Faron, SweetWiki: A semantic wiki, Special Issue of the Journal of Web Semantics on Semantic Web and Web 2.0, Volume 6, Issue 1, February 2008 , Edited by Mark Greaves and Peter Mika, Elsevier, Pages 84-97