Table 4 Confusion matrices for argumentative classification: the first column indeictes the expected category, while the first line provides the measured classification.
From: Gene Ontology density estimation and discourse analysis for automatic GeneRiF extraction
Without sentence positions | ||||
|---|---|---|---|---|
PURPOSE | METHODS | RESULTS | CONCLUS. | |
PURPOSE | 80.65 % | 0 % | 3.23 % | 16 % |
METHODS | 8 % | 78 % | 8 % | 6 % |
RESULTS | 18.58 % | 5.31 % | 52.21 % | 23.89 % |
CONCLUS. | 18.18 % | 0 % | 2.27 % | 79.55 % |
With sentence positions | ||||
PURPOSE | METHODS | RESULTS | CONCLUS. | |
PURPOSE | 93.35 % | 0 % | 3.23 % | 3 % |
METHODS | 3 % | 78 % | 8 % | 6 % |
RESULTS | 12.43 % | 5.31 % | 52.21 % | 13.01 % |
CONCLUS. | 2.27 % | 0 % | 2.27 % | 95.45 % |