Table 6.
Granular test results from model with case features and without lexicon. Scores are over each possible label for the model. Label Count describes how many instances of that particular label is present in the test set, and Prediction Count describes how many predictions the model produces for a particular label.
| Label | F1 | Recall | Precision | Prediction count | Label count |
|---|---|---|---|---|---|
| B-versionEndIncluding | 0.7817 | 0.7817 | 0.7817 | 875 | 875 |
| B-version | 0.8573 | 0.8618 | 0.8527 | 2655 | 2627 |
| B-versionStartIncluding | 0.7415 | 0.7238 | 0.76 | 100 | 105 |
| B-product | 0.8711 | 0.8774 | 0.8649 | 4840 | 4771 |
| O | 0.9935 | 0.9931 | 0.9938 | 184649 | 184768 |
| B-versionEndExcluding | 0.7987 | 0.7922 | 0.8053 | 303 | 308 |
| B-vendor | 0.9126 | 0.8951 | 0.9308 | 2715 | 2823 |
| I-version | 0.4396 | 0.3509 | 0.5882 | 34 | 57 |
| B-versionStartExcluding | 0 | 0 | 0 | 2 | 1 |
| I-product | 0.8549 | 0.8812 | 0.8302 | 3787 | 3568 |
| I-vendor | 0.5714 | 0.5 | 0.6667 | 111 | 148 |
| I-versionEndExcluding | 0 | 0 | 0 | 0 | 1 |
| I-versionEndIncluding | 0.2581 | 0.16 | 0.6667 | 6 | 25 |
| I-versionStartExcluding | 0 | 0 | 0 | 0 | 0 |
| I-versionStartIncluding | 0 | 0 | 0 | 0 | 0 |