CybergeoNetworks 2: a deep learning application for scientific publishing
- Cet article est une traduction de :
- CybergeoNetworks 2 : une application d’apprentissage profond au service de la publication scientifique
- Autre(s) traduction(s) de cet article :
- CybergeoNetworks 2: una robusta aplicación de aprendizaje para publicaciones científicas
Texte intégral
- 1 Pierre-Olivier Chasset, Hadrien Commenges, Clémentine Cottineau, Juste Raimbault
1On the occasion of the conference held on May 26, 2016 to celebrate the 20th anniversary of Cybergeo magazine, four young researchers1 joined forces to develop a pioneering application for geographic, statistical and semantic analysis of data extracted from the Cybergeo corpus (full text and metadata), as well as online data associated with the magazine between 1996 and 2015 (search engines, tweets, citations and publications). The CybergeoNetworks website was born. In keeping with our open-access publishing policy, it provides visual and statistical access to data on the location of the authors of the journal's articles, the countries studied and cited, and who is studying whom? Comprising four interactive modules, the information provided by its analyses is impressive. For example, it is possible to group countries by semantic networks, to visualize the states to which authors are affiliated (information that Lodel, OpenEdition's publishing software, does not provide in its current version), but also the geographical relationships arising from articles published in Cybergeo between an author's state of affiliation and the state(s) studied in the article. In addition to its technical aspects, this innovative tool can also be used to initiate epistemological reflection. A spatial visualization, for example, allows us to represent the closest groups of countries according to the themes associated with their publications in Cybergeo and, beyond that, the words chosen to refer to the country. Algorithmic exploration of publication content and metadata also comes to mind, revealing thematic communities and emerging concepts.
- 2 Article Processing Charge
2In November 2016, a meeting with OpenEdition's R&D department opened up the prospect of a partnership and deployment of the tool to the portal's journals, but other commitments on the OpenEdition platform forced it to give up for a while. Faced with the tidal wave of artificial intelligence, it has become urgent for the French scientific community to equip itself with tools adapted to technological developments, in order to retain control over the future of its research, in a context of surging social networks (ResearchGate, Academia) and commercial publishers in this field and all associated issues (APC2, impacts and bibliometrics).
3This is why Cybergeo, always at the forefront of publishing innovation, is working to set up a new project, CybergeoNetworks 2, in partnership with the Institut des Systèmes Complexes (ISC-PIF) and the CNRS’ InSHS. The aim is to combine CybergeoNetworks modules with Gargantex33, an open-source tool for interactive exploration of large digital corpora. This new application would have the advantage of providing real-time qualitative and statistical analysis of publications and even submissions, accompanied by semantic maps based on data from the various CybergeoNetworks modules.
4In line with the principles of open science, all data, source codes and associated materials will be freely accessible in the Cybergeo Dataverse collection and made available to the community for reuse. At the crossroads of artificial intelligence and scientific work on publication corpora, the application prefigures a new dimension in the self-analysis of research practices, both from an interdisciplinary and international perspective.
Notes
1 Pierre-Olivier Chasset, Hadrien Commenges, Clémentine Cottineau, Juste Raimbault
2 Article Processing Charge
3 https://iscpif.fr/projects/gargantext/
Haut de pagePour citer cet article
Référence électronique
Christine Kosmopoulos, Natacha Aveline et Émilie Lavie, « CybergeoNetworks 2: a deep learning application for scientific publishing », Cybergeo: European Journal of Geography [En ligne], Éditoriaux, mis en ligne le 31 octobre 2023, consulté le 11 février 2025. URL : http://journals.openedition.org/cybergeo/40644
Haut de pageDroits d’auteur
Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.
Haut de page