Classifying patient and professional voice in social media health posts

Table 5 Results for the baseline model trained on all of the training data when testing it on all of test

Train all/test all	Precision	Recall	F1	Support
Other	0.88	0.86	0.87	3358
Patient voice	0.82	0.87	0.85	2395
Professional voice	0.35	0.23	0.28	121
Macro averages	0.68	0.65	0.67	5874
Accuracy	0.85

We report precision, recall and F1 scores per label and overall as macro averages and accuracy as well as the number of test examples (Support)

ISSN: 1472-6947