On September 21, 2021, we have participated in the In-OVIVE seminar. It was an opportunity to discuss our work about representing metadata of meteorological datasets with the In-OVIVE community that has a deep experience in semantic modeling. The slides of our presentation are accessible via the following link. The program of the seminar as well as the complete list of presentations with their slides and videos are available on this link.
On September 1, 2021, our trainees Louis Mendy and Alexandre Champagne have obtained their Msc in computer science after having worked for a period of six months with us. Louis has developed a module to semi-automatically create ontology-based templates (input forms) that facilitates dataset description with semantic data by domain expert. Alexandre has implemented another module dedicated to search datasets based on their metadata. His algorithms implement semantic search techniques based on knowledge bases such as Geonames for spatial search of datasets. Reports of their work and presentations are available on the following links: Louis report, Louis presentation , Alexandre report , Alexandre presentation
On June 30, 2021, we have participated in the IC conference to present our paper "Un modèle sémantique en vue d’améliorer la FAIRisation des données météorologiques" (A semantic model to improve weather data FAIRification). The conference took place between June 30 and July 2 in Bordeaux, France. The program of the conference may be accessed through this link
Abstract: Making meteorological data FAIR in order to ease its reuse is a strategic issue because this data is essential to advance research in many fields. This work proposes a semantic model which combines a metadata model and a data model for describing meteorological observation data. Indeed, modeling (meta)data is an essential step towards their FAIRification. We use the SYNOP open dataset made available by Météo-France to illustrate how difficult data access and understanding can be, and how the use of the proposed model to represent meteorological data improves their compliance with the "F", "I" and "R" principles.
On March 30, 2021, we have given a seminar to Météo-France; we have presented the progress of the Semantics4FAIR project and collected some feedbacks from Météo-France employees as they are data-provider users. Presentation slides are available via this link
On March 29, 2021, Alexandre Champagne has joined the Semantics4FAIR team as an intern for a period of six months to work on the semantic search of datasets.
Internship context : Scientific research communities produce large amounts of data that can be processed in various applications. The Météo-France research center makes part of its datasets available on the Web. However, making the data open and accessible requires considerable effort if one wants to guarantee the quality of the data and their respect of the FAIR principles (Findable, Accessible, Interoperable, and Reusable)[1]. The originality of the approach proposed in this project is twofold: (i) implementation of an ergonomic approach for the collection of users' needs and vocabularies; (ii) a semantic approach to meet the challenge of "Findability".
The semantic approach consists in defining a formal vocabulary (or ontology) that organizes and structures the metadata associated with the datasets, in order to reduce ambiguities when formulating queries to find the datasets that meet a need. Enriching datasets with semantic metadata is a necessary prerequisite to retrieve relevant datasets more quickly before their use by AI algorithms.
The case study of the Semantics4FAIR project concerns meteorological data, which we want to improve the ease of retrieval, including by scientists coming from other domains, such as botanists specialized in pollens. Indeed, the data produced by Météo-France are partly public and open, as all public organizations are committed to do. In spite of this, it remains difficult for a non-specialist to find the data, but also, among all those offered, to select the relevant dataset for a targeted study.
Internship objective : In addition to the work of an ongoing post-doctoral fellow, the internship will contribute to a methodological solution and software to facilitate first the enrichment of MétéoFrance datasets with semantic metadata adapted to a FAIR approach, and then the search of datasets based on these metadata within a portal dedicated to weather data. For this, the internship will include the following tasks: (i) to make an inventory of the databases produced and made available by MéteoFrance, but also to study different portals offering open meteorological data; (ii) to contribute to a semantic representation of the metadata using the business ontology realized by the project; (iii) to implement dataset search functions to facilitate the search of these datasets [2].
[1]: Wilkinson, M. D., Dumontier, M., Aalbersberg, I. J., Appleton, G., Axton, M., Baak, A.,& Mons, B. (2016). The FAIR Guiding Principles for scientific data management and stewardship. Sci Data 3: 160018.
[2]: Chapman, A., Simperl, E., Koesten, L., Konstantinidis, G., Ibáñez, L. D., Kacprzak, E., & Groth, P. (2020). Dataset search: a survey. The VLDB Journal, 29(1), 251-272.
On March 22, 2021, Louis Mendy has joined the Semantics4FAIR team as an intern for a period of six months to work on the specialization of a semantic-metadata management tool.
Internship context : More than one million datasets are published on the web for be exploited/reused by other users [1]. Without metadata documenting their provenance, content, methods used to generate them, etc., their indexing by search engines such as Google Dataset Search, their discovery and reuse are penalized [2][3].
The Semantics4FAIR project aims at improving the accessibility and reuse of datasets produced by MéteoFrance by adding semantic metadata adapted to users. To do so, we have proposed an ontological model that combines standard models for semantic representation of metadata such as GeoDCAT-AP and other vocabularies adapted to meteorological data.
In addition, through another project, we have a software for managing datasets and semantic description of their metadata according to standard vocabularies only (DCAT-AP, OWL-Time, GoeSPARQL, etc.).
However, this software does not allow to add domain ontologies to describe data with domain concepts rather than simple keywords.
Internship objective : The goal of this internship is to evolve this semantic metadata management web application (called DataNoos) to allow it to integrate ontologies specific to the datasets' domains of membership, and to use them to capture these types of metadata. The first model to be integrated will be the Semantics4FAIR ontology describing the metadata of meteorological datasets, and the datasets will be those provided by Météo-France in the framework of the Semantics4FAIR project.
[1]: Benjelloun, O., Chen, S., & Noy, N. (2020, November). Google dataset search by the numbers. In International Semantic Web Conference (pp. 667-682). Springer, Cham.
[2]: Koesten, L., Simperl, E., Blount, T., Kacprzak, E., & Tennison, J. (2020). Everything you always wanted to know about a dataset: Studies in data summarisation. International Journal of Human-Computer Studies, 135, 102367.
[3]: Brickley, D., Burgess, M., & Noy, N. (2019, May). Google Dataset Search: Building a search engine for datasets in an open Web ecosystem. In The World Wide Web Conference (pp. 1365-1375).
We just released the first version of our ontology (semantic model) to represent semantic metadata of meteorological metadata. To ensure the adherence to FAIR principles, our ontology is mainly based on reference ontologies namely GeoDCAT-AP, RDF data cube, and CSVW. Moreover, to explicit semantic entities included in datasets, we have reused domain ontologies such as ENVO, SWEET or AWS. The originality of our work resides in the fact of representing in a fine manner the datasets schema and domain concepts which improves data indexing and discoverability. More details about this version can be accessed on this paper.
Amina ANNANE and Inna KHARCHENKO have joined our team. Amina holds a PhD in computer science; her main research area is the semantic web. She is recruited as a post-doc. Inna is a master student in ergonomics and she is doing her internship with us!
On January 20, 2019, the Semantics4FAIR project members organized a kick-off conference to present the objectives and approach of the project. The conference aimed to share the project issue with other communities interested in meteorological data or concerned about the accessibility of research data, as well as to get feedback. From 13h30 to 17h, several presentations were given according to the following program:
During the second National Open Science Days (JNSO), which took place on November 18-19 in Paris, lessons from the ANR flash call 2019 have been presented (slides). Moreover, each leaureat project leader has presented his/her project by answering three questions: (i) How is the application of open science principles to research data an issue in your field, discipline or specialty? (ii) What are the objectives of the project and the approaches being considered to address them? (iii) What are the prospects in terms of potential applications for the scientific community in the field, other disciplinary fields, or for society?
Answers to these questions in the context of our project Semantics4FAIR, can be found on the following page .
Semantics4FAIR is one of the leaureat projects selected by the French National Research Agency (ANR) in response to the open science flash call 2019 . Please see Section Context for more details about the project.