We should wait for #162 to be finnished. Then we should start to add previously curated data as additional covariates. The purpose of strong the metadata on manually annotated curated data is twofold:
1) As a quality control. No data should be changed in a way that differs compared to manually curated (and checked) data.
2) To store training data from ML models for automated curation. In this way, we can extract "training" data from the corpus.
We should wait for #162 to be finnished. Then we should start to add previously curated data as additional covariates. The purpose of strong the metadata on manually annotated curated data is twofold:
1) As a quality control. No data should be changed in a way that differs compared to manually curated (and checked) data.
2) To store training data from ML models for automated curation. In this way, we can extract "training" data from the corpus.