HoloClean-Legacy-deprecated
HoloClean-Legacy-deprecated copied to clipboard
Inconsistency of domain id in database and the one in feature tensor
If the detected dirty value is null, then in file pruning.py, function _find_dk_domain(), line 202, the candidate of such cell will contain "()". In _create_dataframe() function, line 504 suggests that kij does not record the domain id for value (). But in cooccurrencefeaturizer.py, insert_to_tensor function, line 69, the "()" value in cell domain will have its domain_id, which makes the domain_id in the database and in the tensor not match each other.