« En deçà des outils haut de gamme du Pré-target RAG, par ailleurs travaillés au sein du HN Lab, les briques permettant de déployer une application de découverte de document sur les techniques de la Retrieval-Augmented Generation (RAG) se multiplient fortement depuis quelques mois. Ce billet présente une…
![Knafou, Julien David Marc. «TransBERT: Leveraging Automatic Translation for Domain-Specific Knowledge Transfer » [thèse]](https://i0.wp.com/lalist.inist.fr/wp-content/uploads/2021/03/cnrs_19930001_0279_MD.jpg?fit=1200%2C750&ssl=1)
« Cette thèse vise à surmonter la barrière linguistique en sciences de la vie grâce à TransBERT, un modèle de langue biomédicale entraîné exclusivement sur plus de 22 millions de résumés (abstracts) PubMed/MEDLINE traduits automatiquement en français. Les résultats montrent que TransBERT atteint des performances comparables, voire supérieures, aux modèles de…
« The Empirical Methods for Natural Language Processing Conference (EMNLP 2022) was held both virtually, and in-person, in Abu Dhabi between December 7 to 11. (…) »
« La méthode du crowdsourcing (on dit aussi production participative, externalisation ouverte ou myriadisation) appliquée aux domaines des sciences (on parle alors de sciences participatives, sciences citoyennes ou sciences collaboratives) consiste à faire appel à la foule en vue de réaliser certaines tâches. Si la méthode ne date pas d’aujourd’hui, elle a tout récemment pris une certaine importance…
« The growing deluge of scientific publications demands text analysis tools that can help scientists and policy-makers navigate, forecast and beneficially guide scientific research. Recent advances in natural language understanding driven by deep transformer networks offer new possibilities for mapping science. (…)
Here we report a procedure for encoding scientific…
« Opscidia, a French startup, was the winner in the OpenAIRE open call for innovation. The team set out to build a pipeline of NLP algorithms that automatically generates and represents a domain-specific ontology from text-mining a large corpus of scientific publications covering this domain.
Using…
« More than 50 000 papers have been published about COVID-19 since the beginning of 2020 and several hundred new papers continue to be published every day. This incredible rate of scientific productivity leads to information overload, making it difficult for researchers, clinicians and public health officials to keep up with…
« SDP is a full day workshop that provides an interdisciplinary venue for researchers interested in any aspect of mining scientific literature. (…) »
« (…) The ultimate goal of Prêt-à-LLOD is to create an ecosystem of “ready-to-use” linguistic linked data for natural language processing services and applications to exploit. This in turn aims at promoting the development of a new generation of interoperable services to consume such data. Resources will be linked and accessible by…
« (…) In “Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer”, we present a large-scale empirical survey to determine which transfer learning techniques work best and apply these insights at scale to create a new model that we call the Text-To-Text Transfer Transformer (T5).…
