Semantics of Techno-Social Spaces

Introduction

Nowadays, most of the digital content is generated within public and enterprise techno-social systems like Facebook, Twitter, blogs, wiki systems, and other web-based collaboration and hosting tools, office suites, and project management tools, including Google Docs, SlideShare, Trello and Basecamp. These applications have transformed the Web from a mere document collection into a highly interconnected social space, where documents are actively exchanged, filtered, organized, discussed and edited collaboratively. In these techno-social systems “everything is deeply intertwingled” using the term coined by the pioneer of information technologies Ted Nelson (Nelson, 1974): people are connected to other people and to “non-human agents” such as documents, datasets, analytic tools and concepts. These networks become increasingly multidimensional (Contractor, 2007), providing rich context for understanding the role of particular nodes that represent both people and abstract concepts.

In techno-social systems infrastructures are composed of many layers (such as Internet communication protocols, markup languages, metadata models, knowledge representation languages which have spanned over two decades) and interoperate within a social context that drives their everyday use and development. Abstract concepts become the foci of social interactions. The ability to automatically grasp context and, from context, to infer meaning, becomes very attractive.

It is widely agreed that one of the most important principles in the philosophy of language is Frege’s context principle, which states that words have meaning only in the context of a sentence, that a philosopher should “never ... ask for the meaning of a word in isolation, but only in the context of a proposition” (Frege, 1884). The context principle also figures prominently in the work of Bertrand Russell and Ludwig Wittgenstein. In this chapter we put forward the hypothesis that semantics of the content of techno-social systems should be analysed in the context of the whole techno-social system; not only in the context of one sentence or even the corpus of all textual information in the systems, but based on the whole structure of the techno-social space which includes actors and various artefacts, and the relations between actors and the thing they create and do.

We substantiate our hypothesis by describing a method of formal modelling of techno-social systems and use of graph-based methods for mining such models, and demonstrate applicability of our method for traditional tasks of natural language processing (including term disambiguation and finding semantic foci of a document), as well as for applications in various recommender systems based on a hybrid approach when textual analysis is combined with link analysis.

Nodes in the network represent people, concepts, annotations, projects etc. Links represent social relations (such as friendship, kinship, social roles, etc.), relations in social spaces (such as hasSchool or hasProject), and semantic relations (isA, instanceOf, hasPart, etc.). Socio-semantic relations can be perceived as knowledge in the same way as semantic relations in semantic networks. From the point of view of the traditional dichotomy between top-down and bottom-up approaches to knowledge production and management, network models of techno-social systems can be regarded as bottom-up created social knowledge. When contrasted with ontologies, such networks represent a weaker type of knowledge that lacks conceptualization and cannot be readily used for inferencing. Correspondingly, the potential of this knowledge can be revealed only through the use of robust methods of “soft computing” such as soft clustering and fuzzy inferencing that are tolerant to errors and incompleteness of data, which is endemic in any user-centric knowledge system.