Tag
This paper investigates automated detection of speculative language in biomedical articles using distributed sentence representations, comparing Recursive Neural Tensor Networks and Paragraph Vectors against SVM and other baselines, finding RNTN achieves the best F1 score of 0.885.