Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Table 1.

Examples of sentences from the eDiseases dataset according to their factuality and polarity.

More »

Table 1 Expand

Table 2.

Distribution of sentences into information types (“Facts”, “Experiences” and “Opinions”).

More »

Table 2 Expand

Table 3.

Distribution of sentences into polarity classes (“Positive”, “Negative” and “Neutral”).

More »

Table 3 Expand

Fig 1.

Distribution of sentences into polarity classes (“Positive”, “Negative” and “Neutral”) and factuality classes (“Fact”, “Opinion” and “Experience”).

More »

Fig 1 Expand

Table 4.

Percent inter-annotator agreement for the three polarity labels and the three diseases.

More »

Table 4 Expand

Table 5.

Percent inter-annotator agreement for the three factuality labels and the three diseases.

More »

Table 5 Expand

Table 6.

Feature comparison for the allergies domain.

Results are reported in Accuracy and F-measure. Best results are indicated in bold. For each group of experiments, significance of the best combination of features with respect to the baseline (BoW, W2V and C2V, respectively) is calculated (FDR<0.001***, FDR<0.01**, FDR<0.05*).

More »

Table 6 Expand

Table 7.

Feature comparison for the crohn domain.

Results are reported in Accuracy and F-measure. Best results are indicated in bold. For each group of experiments, significance of the best combination of features with respect to the baseline (BoW, W2V and C2V, respectively) is calculated (FDR<0.001***, FDR<0.01**, FDR<0.05*).

More »

Table 7 Expand

Table 8.

Feature comparison for the breast cancer domain.

Results are reported in Accuracy, F-measure, Precision and Recall. Best results are indicated in bold. For each group of experiments, significance of the best combination of features with respect to the baseline (BoW, W2V and C2V, respectively) is calculated (FDR<0.001***, FDR<0.01**, FDR<0.05*).

More »

Table 8 Expand

Table 9.

Comparison between diseases (allergies, crohn, and breast cancer) (Average accuracy) for the SMO classifier.

More »

Table 9 Expand

Table 10.

Average F-1 by class for the SMO-W2V classifier (Resample).

Best results are indicated in bold.

More »

Table 10 Expand

Table 11.

Classification results when data for the three diseases are combined for the SMO.

Best results are indicated in bold.

More »

Table 11 Expand

Table 12.

F-1 by factuality class and disease for the W2V classifier.

Best results are indicated in bold.

More »

Table 12 Expand

Table 13.

Average F-1 by class for the SMO-W2V classifier (Resample).

Best results are indicated in bold.

More »

Table 13 Expand