Semantics4FAIR ( https://www.irit.fr/semantics4fair/ ) is one of the projects selected by the French National Research Agency (ANR) on theFLASH CALL Open Science entitled "research practices and open research data". It aims at facilitating the tasks of finding and accessing scientific data by a scientific community, in order to support the development of new usages by other scientific communities. The originality of the proposed approach is twofold: (i) a human factor method to capture user’s needs and vocabularies; and (ii) a semantic-based approach takes up the findability challenge thanks to ontologies and metadata templates. We have built an ontology to better describe and document domain specific datasets so that they are more easily findable and reusable byusers from other domains. The core of this ontology reuses standard vocabularies (e.g., GeoDCAT-AP, RDF Data Cube). The domain specific partof the ontology and domain vocabularies account for the various points of view on the domain data and provide a standard language to describethe data sets, their structure and content. To instantiate the developed ontological model by expert domains that are not familiar with ontologies, we have to implement an application with ergonomic interfaces. To tackle this issue, in our project and similar projects like dataNooS, we designed and implemented a prototype of a generic application for semantic dataset description. This application takes as an input an ontological metadata model and allows the user, familiar with ontologies, to semi-automatically define a template (i.e., input form) based on this model. This template allows domain experts to describe datasets with metadata (instantiate the model) without having to deal with ontologies. Once datasets are represented, it is essential to offer search services to improve their findability and reuse. We are working on the development of semantic search algorithms that exploit the ontological model and knowledge bases (e.g. Geonames) to improve the classic facet search of datasets. These applications will be included in the dataNooS platform (see dataNooS poster).This work relies on the collaboration of a computer science lab (IRIT) and a human-factor institute (MSH-T) with scientific communities that want to make their datasets FAIR, and scientific communities that want to reuse this data for their own research projects. We are currently testing the approach with meteorology datasets provided by the atmospheric scientific community (OMP and CNRM), and two data user communities: the Palynologist community (GET) and meteorology data exploitation (Météo-France) services.Scientific Discipline / Research Area: Natural Sciences/Earth and related environmental sciencesRelevance / Link to RDA:We propose a semantic-based approach to improve the FAIRness of scientific datasets in the context of cross-disciplinary research. Wedesigned an ontology and a prototype of dataset description based on this ontology to build a dataset portal. We also designed anexperimental query service where the ontology guides the faceted search for datasets. These contribution and the debates about their design arerelevant for various RDA working groups on metadata FAIRness, such as the following :Link to the Research Data Repository Interoperability WG : FAIR principles and standards adoption; an integrated platform for connecting Open Science eco-system.Link to the FAIR Data Maturity Model WG : a generic platform for domain and community specializations, + Data quality checking.Link to RDA for interdisciplinary research : a generic platform to manage cross-disciplinary datasetslink to Metadata Standards Catalog WG : the generic part of our ontology reuses metadata standardsResearch Metadata Schemas WG : we propose ontology-based templated for research metadata description