Call for Participation

   ECML/PKDD Graph Labelling Workshop  Web Spam Challenge

                          Warsaw, Poland
                       17-21 September 2007

Graph Labelling Workshop

The workshop is focused on the thematic of graph labelling whose goal is to automatically learn to label the nodes of a graph given a known partially labelled graph. The workshop is opened to any submission concerning theoretical models or large size applications of graph labelling with a particular focus on internet graphs.

A special session of the workshop will be devoted to the part 2 of the Web Spam challenge organized with the support of the Pascal Network of Excellence. This challenge focuses on a specific graph labelling problem: detect malicious pages (web spam pages) on a large web graph.

The description of the workshop is given in the Web page :

The description of the Web Spam Challenge is given in the Web page :


Theoretical or experimental research papers representing original, previously unpublished works, are invited to be submitted to the workshop (with a page limit of up to 8 pages). Positioning papers describing current research are also welcome, as well as, opinion papers discussing Graph Labelling models or applications.

Graph labelling is a generic task with several applications in many different domains. The topics include, but are not limited, to:

  • Generic graph labelling models
  • Graph Diffusion
  • Tree annotation models - XML annotation
  • Models for large graphs
  • Web spam detection
  • Wiki, Blog, Web retrieval
  • Web classification and clustering
  • Social networks
  • Collaboration and citation graphs

Web Spam Challenge

Participants to the workshop are also encouraged to participate in the Web Spam Challenge organized within the Pascal Network of Excellence. This competition features a key application, Web Spam detection where the goal is to label the nodes (the Web pages or Web hosts) of a graph as spam or not spam

In order to help ML researchers to participate to this challenge, the organizers of the workshop will provide pre-processed collections. Corpora of different sizes will be made available ranging from small ones (about 1000 nodes) up to large ones (around 1 million nodes). The different collections will be distributed in data formats such as plain text and Matlab data files. This will allow to easily set-up the testing of different types of ML models applied to this task.

Paper submissions

Paper Submissions: The papers must be in English, formatted according to the Springer LNAI guidelines. Papers in this format should not exceed 8 pages.

Important Dates Graph Labelling Workshop :

  • Paper submission : June 30th
  • Paper acceptance notification : July 21st
  • Paper camera ready copy : July 28th
  • Workshop in Warsaw, Poland : Monday, September 17th

Important Dates Web Spam Challenge :

  • Training collections : June 1st
  • Testing collections : July 9th
  • Submission of the results by the participants : July 11th
  • Publication of the results : July 12th
  • Workshop in Warsaw, Poland : Monday, September 17th

Program Committee:

  • Kumar Chellapilla - Microsoft Research
  • Brian D. Davison - Lehigh University
  • Ludovic Denoyer - University of Paris 6
  • Dennis Fetterly - Microsoft
  • Patrick Gallinari - University of Paris 6
  • Remi Gilleron - University Lille 3
  • Marco Gori - DII-University of Siena
  • Mark Herbster - University College London
  • Massimiliano Pontil - University College London
  • Juho Rousu - University of Helsinki
  • John Shawe Taylor - University College London
  • Alessandro Sperduti - University of Padova
  • Tanguy Urvoy - France Telecom Research

Organization Committee:

  • Carlos Castillo - Yahoo! Research
  • Brian D. Davison - Lehigh University
  • Ludovic Denoyer - LIP6 University of Paris 6
  • Patrick Gallinari - LIP6 University of Paris 6

Contact: Ludovic.Denoyer@lip6.Fr