Speaker: Paul Marchal
Date: Thursday 29th of September 2022, 1:20 pm.
Abstract:
Large volume of scientific and technical data have been acquired within research & development projects carried out by Orano and the Université de Lorraine. Throughout this long-term collaboration, top-ranked articles have been published based on genuine datasets resulting from specialized analyses, in particular mineralogy, isotope geochemistry and fluid composition. Samples are stored in a well referenced lithothèque and regularly used for new batches of analysis. These results have largely contributed to the knowledge of the unconformity related deposits and are considered as key parameters for defining new projects, constraining any modelling and defining machine learning analysis strategies. In order to provide access to these numerical datasets through Otelo and to connect to the Orano standards under acQuire, a joint-project for data valorization has been set up in order to develop semi-automated methods for historical data integration and design of database templates. This project aims first to provide standards methods and templates to help new data integration into Otelo’s data repository and secondly to build tools to format and integrate historical datasets. The application of these methods and standards contribute to making datasets easier to find, accessible, interoperable and reusable (FAIR Principles). For dataset, we focused especially on drillhole data analysis, i.e. data associated with sampling on cores. Concerning new data integration, initial work was the redaction of good practice guidelines1 for data management based with focus on: Explicit & coherent data naming; access & centralization of all the necessary data; use of international standards; minimization of useless duplicates & blanks. To implement, we elaborated a standardized data template containing all required metadata data fields, in order to generalize for each new project, the realization of an integrable data file collecting all samples analysis data. Then, based on established models and templates, we designed a JavaScript web-application, on the one hand to help new file edition & also to assess historical data files quality and on the other hand extract information from sparse files. We had also implemented a geographical visualization tool based on drillholes. It allows to compare data from different projects & to extract samples lists according to available attributes, either it’s about project metadata or analysis data values.