The CEA List institute achieves technology transfer in the field of smart digital systems. The team involved works both in computer vision and computational linguistics, relying on machine learning. In particular, it works on Entity Linking and cross-modal retrieval using textual-visual joint spaces . It has a strong experience in evaluation campaign organization (e.g Retrieving Diverse Social Images task at Media-Eval from 2013 to 2017) and participation.
Persons: Omar Adjali, Romaric Besançon, Olivier Ferret and Hervé Le Borgne.
INRIA participates with team LinkMedia, focusing on content-based media linking and aiming to create explicit links at different levels to reflect context: at signal level, e.g. repeating patterns, or at a semantic level, e.g. to follow topics or stories. The team has in particular a strong experience in
computer vision. It has pioneered research in large-scale visual retrieval and nearest neighbor search and holds state of the art with conventional features and more recent deep visual representations, including manifold learning.
Persons: Laurent Amsaleg, Yannis Avrithis and Ewa Kijak, Shashank Venkataramanan
IRIT represents one of the most important potential in computer science research in Southern France. The 24 research teams of the laboratory are structured into seven scientific topics that cover all the fields of computing sciences. Jose Moreno and Yoann Pitarch are associate professors and Lynda Tamine is a full professor, all them members of the IRIS team. Their more related work to the project focuses on textual-graph representation and entity embedding, and their applications.
Persons: Jesús Lovón, Jose Moreno, Yoann Pitarch and Lynda Tamine
LISN was created on january 2021, integrating the LRI and LIMSI. This last has a strong expertise in natural language processing. Anne Vilnat, professor, Sahar Ghannay, assistant professor, and Anne-Laure Ligozat, associate professor, are members of the ILES group and work on entity linking, representation learning, information extraction and question answering from textual data. Camille Guinaudeau, associate professor from the TLP group, is interested in person identification in multimedia documents and information retrieval in multimodal data.
Persons: Sahar Ghannay, Camille Guinaudeau, Paul Lerner, Anne-Laure Ligozat and Anne Vilnat