Introduction - Social Semantics: The Search for Meaning on the Web - page 6

A much shorter version of this work has been previously published as Sense and
Reference on the Web in the journal Minds and Machines (Halpin 2011).
Then we dive from the heights of theory to the depths of experimental work. In
Chap. 5 we begin the exploration of an alternative form of discovering the meaning
of a representation, namely that of 'bottom-up' collaborative tagging systems, where
users simply 'tag' a resource with a term they find useful or descriptive and so
define the 'sense' of a URI as a set of terms. We commit a number of experiments
to determine if these tags converge over time and over a diversity of resources.
Then in Chap. 6 we extend this exploration to search engines, considering the 'bag-
of-words' produced by a document to be equivalent to a set of tags, and therefore
the sense of the URI. In particular, we explore this using documents from both
the Semantic Web and the hypertext Web, and use relevance models to combine
them. The study of tagging was previously published as The Complex Dynamics of
Collaborative Tagging in ACM Transactions on the Web co-authored with Valentin
Robu and Hana Shepard (Halpin et al. 2007; Robu et al. 2009), while the user study
was co-authored with Dirk Bollen as An Experimental Analysis of Suggestions in
Collaborative Tagging (Bollen and Halpin 2009). A few elements of the study
of search engines and relevance feedback was previously published as Relevance
Feedback between Web Search and the Semantic Web with Victor Lavrenko, who
co-wrote some of the text and the equations (Halpin and Lavrenko 2011b) with a
longer version published as a journal article entitled Relevance feedback between
hypertext and Semantic Web search: Frameworks and Evaluation in the Journal of
Web Semantics (Halpin and Lavrenko 2011a).
We finally turn to formulate a third position in Chap. 7, social semantics ,which
states that since the Web is a form of language, and as language exists as a public
mechanism among multiple agents, then the meaning of a URI is determined by the
socially-grounded use of networks of representations on the Web by ordinary users .
As vague as this position seems at first glance, we argue this analysis of meaning
and representation is the best fit to how natural language works, and it supersedes
and even subsumes the two other positions. Furthermore, it goes beyond a certain
quietism about natural language attributed to Wittgenstein as well as a certain belief
in the occult powers of some 'mental' lexicon. Ideas in this version were previously
published with Andy Clark and Michael Wheeler as Towards a Philosophy of
the Web: Representation, Enaction, Collective Intelligence (Halpin et al. 2010).
The entire Ph.D. thesis was submitted and approved to University of Edinburgh,
with Yorick Wilks being the external examiner, as Sense and Reference on the
We b (Halpin 2009b), with the precis being published with Henry S. Thompson
as Social Meaning on the Web: From Wittgenstein to Search Engines in IEEE
Intelligent Systems (Halpin and Thompson 2009).
As Wittgenstein would say, one must remember that every “language game”
comes with a “form of life” (Wittgenstein 1953), and the Web is a new form of
life that goes beyond the philosophy of natural language, and leads us straight into
a new philosophy of dynamic machinic and human assemblages, a philosophy-to-
come of collective intelligence.