Social Websites – Complex Dynamics and Structure

a satellite conference of ECCS'07

October 5th, 2007


9.30-9.40 opening
9.40-10.40 Gerd Stumme (University of Kassel)
Information Retrieval and Knowledge Discovery in Folksonomies
Social bookmarking systems allow users to organize and share bookmarks, photos, etc. on the web. The reason for the success of these systems lies mainly in the fact that no specific skills are needed for publishing and editing. As these systems grow larger, however, the users feel the need for more structure for better organizing their resources. This, together with the new underlying data structure – so-called folksonomies – provides new challenges for the research areas of information retrieval and knowledge discovery. In the talk, I will first present our social bookmarking system BibSonomy, which supports the scientific community in organizing and sharing bookmarks and lists of publication references; and which serves as a platform for our experiments with IR and KDD algorithms. In the second part of the talk, I will present selected research results, including a ranking algorithm and a triadic clustering algorithm for folksonomies.
10.40-11.00 coffe break
11.00-11.45 Riley Crane (ETH Z├╝rich)
A Shocking Look At YouTube
We present a comprehensive study of the dynamics of the daily view count for nearly 5 million videos on the popular video-sharing website The processes governing the dynamics of videos is remarkably regular – from the Poisson process governing the uploading of videos to the processes governing the relaxation dynamics following a shock in the daily view count. While most videos disappear into obscurity never receiving more than several views, we find a large subset (10%) which experience a shock. We will explore the origin and nature of these shocks and their subsequent relaxation towards the pre-shock state.
11.45-12.30 Ciro Cattuto (Centro Studi e Ricerche “Enrico Fermi”)
Memory and Growth in Collaborative Tagging Systems
Collaborative tagging systems are attracting the attention of researchers in complex systems because they expose interactions between users and the information they process, as well as interactions among the users themselves. We report on some stylized facts observed across different systems, and – based on a large-scale snapshot of a popular system – we show that the growth over time of the tag vocabulary seems to follow a universal behavior. We also report data on the frequencies of usage of tags, investigate correlations in tag streams, and show that a simple stochastic model with heavy-tailed memory can reproduce the experimental data.
12.30-14.15 lunch break
14.15-15.15 Ulrik Brandes (University of Konstanz)
Visual Analysis of Controversy in User-generated Encyclopedias
Wikipedia is a large and rapidly growing Web-based collaborative authoring environment, where anyone on the Internet can create, modify, and delete pages about encyclopedic topics. A remarkable property of some Wikipedia pages is that they are written by up to thousands of authors who may have contradicting opinions. We define a network of authors revising other authors and present methods for its visual analysis that yield interesting insight into the structures of controversy by highlighting the dominant authors of a page, the roles they play, and the alters they confront.
15.15-16.00 Vito D. P. Servedio (University of Roma “La Sapienza”)
Network Structure of Folksonomies
Folksonomies can be viewed as three mode graphs or as graphs made up of nodes (tags, users, resources) connected by hyper-edges. I shall report on some network statistical properties of a folksonomy graph based on data collected for the system. Moreover, by introducing a suitable distance between resources based on tag co-occurrence, I shall show that folksonomies embed a meaningful semantic clusterization of resources.
16.00-16.30 coffee break
16.30-17.15 Filippo Menczer (Indiana University and ISI Foundation)
Social Search
This talk will present two research projects under way in the Network and agents Network (NaN), which study ways of leveraging online social behavior for better Web search. is a social bookmarking site where users donate their personal bookmarks. A search and recommendation engine is built from a similarity network derived from the hierarchical structure of bookmarks, aggregated across users. 6S is a distributed Web search engine based on an adaptive peer network. By learning about each other, peers can route queries through the network to efficiently reach knowledgeable nodes. The resulting peer network structures itself as a small world that uncovers semantic communities and outperforms centralized search engines.
17.15-17.45 Bettina Hoser (University of Karlsruhe)
Linked? – Multiple link relationships in Social Sites
The growing interest in social sites in scientific research as well as in business has led to a sharp increase in the analysis of such sites. The tools and methods to technically analyze such sites are developed and are being enhanced. One major question though seems to be left mostly unattended: If one takes the relationship data in a site like e.g. MySpace, what is the ‘real’ interpretation of the social network analysis? There is no clear definition as to the meaning of a link. Is it frienship, acquaintance, awareness? When analyzing the dynamics of such a site, the question of time dependent relevance of links is added to this problem. In this talk these points will be adressed with the goal of bringing into the focus of the discussion the content driven interpretation of the analysis of social sites.
17.45-18.00 discussion and wrap-up


Participation is free of charge, but requires registration to the main ECCS07 conference (early registration deadline is May 15th). To register for ECCS07, please visit the official ECCS07 registration page and specify your interest in this satellite conference. We would appreciate if you could also send email to indicating your intention to participate.


For any inquiries about the conference, please send email to


A new kind of user-centered applications such as folksonomies, recommendation systems, wikis and so on, now commonly referred to as "Web 2.0" have gained considerable ground in the past 2 years. As examples we can point to Flickr,, YouTube, BibSonomy, and many more. In addition, the economic impact of these websites is surging, as can be readily seen from the fact that Yahoo acquired Flickr, News Corp bought MySpace and OpenBC went public in the last few months.

The collaborative character underlying many Web 2.0 applications puts them in the spotlight of complex systems science. Modern web applications for organizing and sharing data are open-ended, dynamically evolving, collaborative and therefore intrinsically social. The huge popularity acquired by collaborative tagging systems, for example, shows that the computer-mediated interaction of web users can create valuable and complex information architectures. These structures are genuinely emergent and early research is reporting rich and often surprising features. The problem of linking the low-level scale of user behavior with the high-level scale of global applicative goals is a typical problem tackled by the science of complexity: understanding how an observed emergent structure arises from the activity and interaction of many globally uncoordinated agents. At the same time, from the viewpoint of Information Technology, this appears as an engineering problem, and its solution lies in understanding and eventually controlling the self-organized structures that arise in modern web-based systems.

The large number of users involved, coupled with the fact that their activity is occurring on the web, provide for the first time a unique opportunity to monitor the “microscopic” behavior of users and link it to the high-level dynamics of applications, by using formal tools and concepts from the science of complexity.

Aims and Goals

Our aim is to leverage the large popularity and scale of the ECCS’07 conference to bring together researchers from Physics, the Semantic Web community, the Social Network Analysis community, the Web 2.0 community and other research areas. We plan to increase collaboration and exchange of experiences, with a concrete focus on mathematical tools and models, experimental data, and techniques for data analysis.

Our goal is also to start creating a network of researchers with a common base of inter-disciplinary knowledge, as the basis of a new research community.

Finally, this research field is very young and there are only a few European players. We would like them to become aware of each other, also in view of the call for new projects in FP7, for which the topics of this conference will be of high priority in the IST call.


  • Bettina Hoser, University of Karlsruhe (TH), Germany
  • Ciro Cattuto, University of Rome "La Sapienza" of Rome, Italy