Dataset as a tree/graph. If you can imagine that a dataset will undergo changes like normalization and/or context changes, it needs to be presented as such with transparency for the user and agents. Also it can be across different domains (such as how cs-analyze would generate new datasets, it can then be imported into this management system. ) so a selection system is needed, since this lean ecoxai is only for a single dataset
A dataset should be considered one or more ingestible data. the user context/research question should be on that dataset, however complex
Current implementation:
single dataset upload (doesn't handle multiple files for a single dataset yet)
single folder (/backend/datasets)
Future features:
multiple datasets selection (updates in the sql and workflow for data and datasets folder needed)
backend assets management
versioning
integration into the hypothesis graph as bonus
Dataset as a tree/graph. If you can imagine that a dataset will undergo changes like normalization and/or context changes, it needs to be presented as such with transparency for the user and agents. Also it can be across different domains (such as how cs-analyze would generate new datasets, it can then be imported into this management system. ) so a selection system is needed, since this lean ecoxai is only for a single dataset
A dataset should be considered one or more ingestible data. the user context/research question should be on that dataset, however complex
Current implementation:
single dataset upload (doesn't handle multiple files for a single dataset yet)
single folder (/backend/datasets)
Future features:
multiple datasets selection (updates in the sql and workflow for data and datasets folder needed)
backend assets management
versioning
integration into the hypothesis graph as bonus