Table 4:
System | Direct Edits |
Guides Edits |
Generalizes Edits |
Limitations with respect to Data Completion |
---|---|---|---|---|
Falcon | ✓ | × | ✓ | Does not guide the user on edits. |
SampleClean | × | × | ✓ | Addresses duplication and value errors for aggregate queries only. Does not fill in incomplete datasets. |
Data Imputation & Holoclean | × | × | × | Low precision or recall, as shown in Section 4.4 and 4.6.3, since limited evidence in the data for missing values. |
Transformational Edits -Trifacta, Potter’s Wheel, Polaris | ✓ | × | ✓ | While these systems generalize edits based on transforms, they do not guide users on effective transforms. |
Interactive Learning - ActiveClean, Guided Data Repair | × | ✓ | ✓ | Suggests rules based on underlying models, which is not applicable when specific data instances are missing. |