From April 11 to 13, the fifth edition of IAPR TC15ís workshop on Graph-based Representations was held in Poitiers, France.† From previous editions, we were already accustomed to a high quality of presented works, accompanied by a warm, friendly atmosphere favoring discussion and exchange of experiences and information.† Thanks to the invaluable efforts by Luc Brun, this edition was wonderfully aligned to the high standards set by the past GbR workshops, even if the number of submitted papers significantly increased with respect to previous editions.†
In order to accept all the deserving papers without loosing the "feeling" of the workshop by introducing parallel sessions or increasing the number of days, we experimented with having some authors present a very short oral introduction of their papers followed by a poster session where more details were provided.† This experiment worked quite well, and this way of presenting the papers helped to keep the attention and the focus of the participants during the whole session.
Another innovation this year was the joint organization with the workshop on Discrete Geometry for Computer Imagery (see related article on DGCI 2005).† One of the three days of GbR was devoted to a joint session with DGCI, with papers covering the borderline between the interests of the two communities.† The success of this joint organization, which has drawn the interest of some DGCI members to GbR activities (and vice versa!), has induced us to decide to repeat this format for the next GbR edition, obviously choosing a different peer workshop.† For this reason, we have not yet established the location of the next edition, waiting to receive proposals involving other workshops that will be held in 2007.† Some interesting candidates have already been proposed.
Within the discussion panel on the future of GbR, the issue of creating a repository of datasets and algorithm implementations was discussed.† This repository would help make the results of our research reproducible and comparable with other approaches.† It has been suggested that in the future the reviewers of the papers submitted to our workshop should invite the authors, where it makes sense, to either use the datasets already in the repository for their experiments, or to share their own datasets through the repository.† A few volunteers have given their availability to implement a first working version of the repository, limited to the very basic functionalities but sufficiently complete to allow us to start gathering data.†
The repository, as was pointed out during the discussion, could become an effective resource to make our activities more widely known outside of the TC15 and IAPR communities.†
Other suggestions that emerged for promoting the visibility of TC15 was a tighter coordination with other technical committees (in order to avoid the useless duplication of common efforts) and the organization of a special session within larger conferences, even outside of the IAPR community.† In particular, it was felt we should aim at reaching the research communities of Asian countries, which up to now have had only a marginal participation to our activities.
Also, it was suggested that we should exploit the proposal of Elsevier to publish textbooks on the subjects covered by IAPR; in particular we should try to use this opportunity to realize textbooks covering both the basic and the advanced aspects involved in the use of graph based representations.
For more information on IAPR TC15 Graph-based Representations see