Co-auteur
  • VENTURINI Tommaso (5)
  • GIRARD Paul (4)
  • OOGHE Benjamin (4)
  • JACOMY Mathieu (4)
  • Voir plus
Type de Document
  • Article (3)
  • Communication non publiée (3)
  • Site Web (3)
  • Communication dans des actes de colloque publiés (1)
in Actes des 23èmes Journées francophones d'Ingénierie des Connaissances (IC 2012) Publié en 2012-06-25
DECLERCK Gunnar
AIMÉ Xavier
CHARLET Jean
2
vues

0
téléchargements
Ce texte se propose de discuter l'idée que les ontologies fondationnelles sont utiles, voire nécessaires, au bon fonctionnement des systèmes de traitement de contenus, en particulier à leur interopérabilité sémantique. Après un rappel des principales caractéristiques des ontologies, nous proposerons une liste des grandes fonctions aujourd'hui attribuées ou au moins attendues des OF, pour en discuter ensuite le bien-fondé. Nous montrerons que la possibilité de réaliser une ontologie intégrant des primitives et définitions formelles suffisamment générales et génériques (universelles) pour décrire la sémantique des concepts de domaines spécialisés du savoir est loin d'être acquise ; et que (ii) même si une telle ontologie s'avérait réalisable, il n'est pas certain qu'elle permettrait d'assurer l'IS entre systèmes, c'est-à-dire un échange de données en préservant le sens.

The aim of the project MEDEA is to use digital mapping tools to try and trace the provenance and pivot points in these debates within different institutional settings. We look principally at the construction of scientific expertise in the Intergovernmental Panel on Climate Change (IPCC) and the evolution of adaptation as an issue within the international negotiations of the United Nations Framework Convention on Climate Change (UNFCCC). We also have a separate case study that follows regional deliberations on climate impacts and water management in the southwest of France. This website presents the results from the first two investigations of this ANR funded project, which was was made possible by a joint collaboration between social and data scientists (Sciences Po, médialab), climate scientists (Laboratoire des Sciences du Climat et l'Environnement) and designers (Ecole Nationale Supérieure des Arts Décoratifs). Experimental in its combination of disciplines and digital data methods, our mapping activities attempt to make legible to a non-expert public the institutional and topical transformations within the leading scientific and political arenas of the climate debate.

Cet article reprend une recherche de Luc Boltanski sur les enseignants de l’IEP de Paris. Dans cette recherche, Boltanski s’appuie sur une représentation tabulaire des champs sociaux pour montrer que la classe dominante se caractérise avant tout par sa multipositionnalité, c’est-à-dire par la tendance de ses membres à occuper plusieurs positions dans plusieurs champs. En remplaçant le tableau de Boltanski par un graphe d’individus et d’institutions, nous discuterons les caractéristiques et les avantages d’une sociologie de réseaux hétérogènes.

Since its foundation in May 2009, Sciences Po’s médialab has worked to enhance the use of digital methods and tools in Social Sciences. With the help of current tools and methods, we experienced the use of web mining techniques to extract and mine digital traces (hypertext links, spontaneous expression on blogs or social networks...) of collective phenomena. Our intention is to consider the web as a field to build new kind of corpora, and not as a research object in itself (web studies), neither as a media (innovative digital mediated surveys) nor as a medium (publishing or accessing structured digital data from the web). This approach raised methodological and practical issues starting with the difficulty to build the highly accurate corpora needed by social scientists from the very complex document space that is the web : it has no size (too big, too dynamic), no clear boundaries because of its hyperlink structure and is composed of a wide heterogeneity of documents (technically, in usage, in time). How to qualitatively identify, select and collect web resources in such a quantitative context ? What does accuracy and representativity means in the moving matters of the web ? What are the tools which can equip the social scientists to build those new kind of corpora ? Because we couldn't find a good enough answer to those questions by using the existing tools we decided to launch in October 2010 the Hypertext Corpus Initiative gathering actors from web archiving, web mining, social sciences and librarians communities. HCI provides for social scientists a new set of methodology and tools, allowing them to mine more accurately digital traces of social phenomena from the web. We will present in this paper the 4 mains methodological and technical issues discussed in HCI which lead us into developing a new set of tools : (1) “what is a web corpus ?”, introducing the concept of web entities to handle the complexity and heterogeneity of web resources; (2) “how to build a web corpus ?”, the methodological and technological issues regarding the quali- quantitative process of building a web corpus proposing to organize a research driven crawling for social sciences purposes; (3) “how to analyze a web corpus ?”, we would like to identified opportunities and limitations in using the web as a research field; (4) “how to foster the use of web archives by social scientists ?”, by applying web corpus principles to the archived web.

Publié en 2014-10
MEUNIER Axel
JACOMY Alexis
PLIQUE Guillaume
GUIDO Daniele
BAYA LAFFITE Nicolas
ROGERS Richard
MUNK Anders Kristian
BORRA Erik
BOUNEGRU Liliana
CIUCCARELLI Paolo
KALTENBRUNNER Andreas
LANIADO David
5
vues

0
téléchargements
This website presents the results of the EU research project EMAPS, as well as its process: an experiment to use computation and visualization to harness the increasing availability of digital data and mobilize it for public debate. To do so, EMAPS gathered a team of social and data scientists, climate experts and information designers. It also reached out beyond the walls of Academia and engaged with the actors of the climate debate.

Publié en 2015-10
PEDROJA Cynthia
ANSART Thomas
MITRANO Patrice
MUNK Anders Kristian
6
vues

0
téléchargements
Les images du médialab et de l'atelier de cartographie se rencontrent et dialoguent dans les vitrines de la bibliothèque. Retrouvez les textes et téléchargez les images qui vous ont marqué.

in Bulletin de l'Association Française pour l'Intelligence Artificielle Publié en 2011
BRINGAY Sandra
BRICON-SOUF Nathalie
2
vues

0
téléchargements

in Bulletin de l'Association Française pour l'Intelligence Artificielle Publié en 2011-10
BRICON-SOUF Nathalie
BRINGAY Sandra
1
vues

0
téléchargements

Khartis results from an observation: how can we create easily a statistical map, with a few clicks, without loosing the scientific accuracy and rigor while processing the data? Khartis is not the first mapping tool available online! It is, however, one of the few to propose different perspectives of the world representation among which the cartographer will choose. In order to feed the tool with data, you may want to retrieve from the Khartis website the data used to produce the maps given as examples or to import or copy / paste your own datasets. The base maps are geo-referenced and allow you to work on the proposed entities (countries, regions...) but also on objects identified by pairs of coordinates. You can choose between several projections (besides Mercator) and configure the visual appearance of your data.

Suivant