Pragmatics and semantics in social tagging systems

Social tagging systems, such as delicious and flickr, have emerged as participative and scalable alternatives to traditional forms of information organization. In social tagging systems, users tag resources (such as URLs or pictures) with tags for e.g. later retrieval or re-finding. The structures emerging in such systems (e.g. the latent semantic relationships between tags) have been referred to as "Folksonomies", and are in the focus of recent and current research efforts that aim to understand the role of tagging systems in ontology learning, knowledge acquisition and information retrieval. However, the current state of research offers little understanding about how these structures emerge in social tagging systems, and to what extent they are influenced by user behavior (i.e. the pragmatics of tagging). Conversely, we know little about how user motivation and behavior is influenced by emergent structures in social tagging systems. This project is concerned with the systematic exploration of mutual interactions between user behavior and motivation on one hand, and the emerging structures that can be observed in social tagging systems on the other. The first part of this project focuses on the development of methods for the automated extraction of semantics from social tagging systems, also known as "ontology learning" methods (tag semantics). The second part focuses on analyzing users who are collecting information in social tagging systems, and their motivations and behavior (tagging pragmatics). Analyzing user behavior will yield novel insights into how and why users use social tagging systems, and what goals they pursue by using them. The third part of this proposal focuses on the combination of the two aforementioned problems, and tries to relate knowledge about user behavior with knowledge about emergent semantic structures, to better understand mutual interactions and influences between these two perspectives. Based on preparatory work on emergent semantic structures of tags (Andreas Hotho, Würzburg) and user motivation and behavior in tagging systems (Markus Strohmaier, Graz), this project aims to identify and investigate mutual influences between those two problems. The combination of the participant`s research background together with the availability of a social tagging system that allows "living lab" experiments (http://bibsonomy.org, maintained by Andreas Hotho) puts this bilateral project in a unique position to address the defined research questions by conducting "in the wild" studies involving A/B testing or controlled experiments with live users. The overall contribution of this project is a better understanding about interactions between user behavior and emergent structures in social tagging systems, and opportunities to influence properties of social tagging systems.

The goal of this project was the study of pragmatics and semantics in collaborative information systems which have become a central aspect of todays Web. They represent a participative and scalable alternative compared to traditional forms of information organization. Through participation in such systems, structure emerges that has been of interest for researchers for some years now. Pragmatics and semantics so far have mostly been studied separately; in this project the goal was to study the mutual influences between them. Both tagging systems as well as Wikipedia were analyzed regarding pragmatic and semantic aspects. This project provides an extension to the array of methods available for understanding user motivation on the Web as well as a new view on pragmatics by studying human navigational behavior on the Web. This has also lead to the analysis of the usefulness of human navigational paths for the task of determining semantic relatedness scores between concepts. The application of the methods developed in this project aim at providing new insights into human behavior in collaborative information systems as well as into the discovery of semantic structures on the social web.