[AI Chair] Interview with Claire Gardent
23 July 2020
In December 2019, 40 French projects out of a total of 173 were awarded “Research and Teaching Chairs in Artificial Intelligence” following a French National Research Agency (ANR) call for projects. Two of these projects are from the Lorraine region. Zoom in on Claire Gardent, CNRS Research Director at Loria, one of the selected projects’ coordinators.
What is your AI Chair’s research theme?
Claire Gardent: “The XNLG Chair focuses on Natural Language Generation (NLG). In very general terms, Natural Language Processing (NLP) models either the comprehension of texts (going from text to meaning) or their production (going from meaning to text). In XNLG, we will be studying the development of neural methods for the generation of texts in multiple languages (most current work focuses on English) and from different sources – semantic representations, databases, knowledge bases and text. Generating texts from databases or knowledge bases means that these bases can be visualized in natural language which is accessible for everyone. Generating text from text means that text can be summarised, simplified or paraphrased.”
What are the possible applications in real terms?
Claire Gardent: “XNLG is a fundamental research project. Eventually, the applications of this research will include database or knowledge base verbalization, automatic summarisation , simp and text simplification or paraphrasing”.
What does obtaining this Chair mean to you and what will it enable you to do?
Claire Gardent: “In the last five years, research in the field of NLP has experienced a veritable growth explosion. There were around 500 submissions to major conferences in the field between 2005 and 2015. These have tripled in the space of five years with 1740 submissions to the 2019 ACL* conference. Private companies have set up research centers involving the best researchers working with powerful computing architectures and large quantities of textual data. Neural methods are evolving extremely rapidly. In this highly competitive context, XNLG will strenghten the existing research group within the Loria’s** Synalp team which works on Natural Language Generation. The four early stage researchers who will be recruited in the framework of the XNLG AI Chair will help strengthen and develop research already ongoing in the group as part of the ANR Quantum project on automatic question generation, of the innovative training network NL4XAI (Natural Language for Explainable AI, a H2020 project) which works on developing interpretable models for text generation and in the framework of five other PhD theses on text generation for dialogue, fact checking and verbalisation of knowledge bases. More generally, the Chair obtained for the XNLG project will enable the development of neural methods to generate multilingual, multi-source texts and will provide training for young high-level researchers in this field.”
*ACL: Association for Computational Linguistics
**Loria: the Lorraine Research Laboratory in Computer Science and its Applications is a joint research unit (UMR 7503) common to the CNRS, the French national research institute for digital science and technology (INRIA) and the Université de Lorraine.
This article was published on 25/02/2020 in factuel.univ-lorraine.fr.