MLentory is centered around information on ML models, how to harmonize that data, and how to make it available and searchable on an FDO (FAIR Digital Object) registry.

MLentory is providing an FDO (FAIR digital object) registry. It aggregates metadata on ML models from various sources with a search engine. A recommendation service makes it easier to find ML models and models similar to a reference model (e.g., a model selected by the end user).

MLentory follows an Extraction, Transformation and Loading approach (ETL) where information is extracted from the sources and stored in temporary CSVs that are later transformed to a common metadata schema. Harmonized metadata is stored in a triple storage and indexed. FDOs are supported with Signposting on the landing pages to RO-crates used to package the metadata. A REST-like API is also provided.

Target research artifacts are Metadata on ML models. Outputs are harmonized and aggregated metadata on ML models. They are available via user interface, triple store (SPARQL and direct download) and API.

This service is currently under development.