📑Datasets

This use case consists in allowing the pooling of aggregated data in order to train artificial intelligence algorithms. This is a necessity for the public and private sectors, as well as for research.

This use case would allow to answer needs such as:

• To cross-reference the usage data of digital readers and build a model that would allow to detect school dropout very early, or on the contrary an interest that is not reflected in the assessments

• Facilitate the development of adaptive learning solutions, based on external data

• Impact studies by cross-referencing data from individuals throughout their lives (school, orientation, training, skills, career)

Benefits

- Allows actors to cross data sets, for instance learning traces, that are currently fragmented

- Limits cold start problems

- Exploit unused data in a trusted environment

- Larger datasets to open up training possibilities for Machine Learning models

- Interoperability of educational data

Building blocks mobilized

• Contract: which lowers the legal and contractual barrier and makes it accessible to actors of all sizes

• Learning traces interop and skills data interop: to build a coherent dataset from heterogeneous data by source and format

• Anonymization and pseudonymization: secure the supplier, allowing him to provide a dataset compatible with the regulations

• Consent: easily integrated with the provider's services. The dataset produced is generated by users who are informed of the purpose and have given their consent.

Among its early adopters and data providers,Prometheus -X brings together a set of stakeholders committed to implementing this use case.

These organizations are: Pôle Emploi, INRIA, FUN MOOC, Insititut Mines Telecom, Université de Lille, MENJS, Openclassrooms, WebForce 3, Serious Factory, Weenoz, Digischool, numerous edtechs, ...