Table 5 Evaluation results using the Geoquery250 corpus
From: Comparative study on the customization of natural language interfaces to databases
NLIDB | Total queries | Queries with substitutions | Correctly answered | Recall (including substitutions) (%) | Recall (excluding substitutions) (%) |
|---|---|---|---|---|---|
Our NLIDB | 250 | 0 | 141 | 56.4 | 56.4 |
ELF | 250 | 0 | 89 | 35.6 | 35.6 |
C-Phrase | 250 | 63 (38 correct) | 179 | 71.6 | 56.4 |