This new feature would ask the user for more information or data if insufficient in the {dataset_description, causal_query, dataset_file}. This can be in the form:
(1) If the variable names in the CSV file are not defined in the dataset description, ask the user to define each of the column names. We can have a separate flag to ignore this during experimental runs, since everything is well-defined in CauSciBench
(2) If there is additional data needed for identification purposes, ask the user. This is a harder problem, will need to think about this carefully.
This new feature would ask the user for more information or data if insufficient in the {dataset_description, causal_query, dataset_file}. This can be in the form:
(1) If the variable names in the CSV file are not defined in the dataset description, ask the user to define each of the column names. We can have a separate flag to ignore this during experimental runs, since everything is well-defined in CauSciBench
(2) If there is additional data needed for identification purposes, ask the user. This is a harder problem, will need to think about this carefully.