Demonstrating Data-to-Knowledge Pipelines for Connecting Production Sites in the World Wide Lab: Trajectory Data and Benchmark Models

These are the trajectory data created as part of the Demonstrating Data-to-Knowledge Pipelines for Connecting Production Sites in the World Wide Lab Project: https://s.fhg.de/gorissen-2025a.One csv file is one trajectory. For implementation details, check the project website for the code repository link. In the folder test you can find data used for model evaluation or testing. Metadata must be derived from the metadata_dump_test.json. In the folder train you can find data used for model training and cross validation. Here, the data is split into subfolder representing the instance that created the data. Further, metadata must be derived from the metadata_dump_train.json.f2e72889-c140-4397-809f-fba1b892f17a: LLTc9ff52e1-1733-4829-a209-ebd1586a8697: ITA2e60a671-dcc3-4a36-9734-a239c899b57d: WZL The files metadata_dump_*.json provide a complete metadata dump for each folder, as metadata is stored in a FileObject on coscine, rather than as a file. That is to say, these files are generated for publication completeness. These are the benchmark models created as part of the Demonstrating Data-to-Knowledge Pipelines for Connecting Production Sites in the World Wide Lab Project: https://s.fhg.de/gorissen-2025a. Models are Tensorflow Models. For implementation details, check the project website for the code repository link. In the folder pre-trained you can find both the foundation and instance specific used for transfer learning approaches in the benchmark, as well as associated hyperparameters.