Detecting Speculative Language in Biomedical Texts using Recurrent Neural Tensor Networks

In three linesDetection of speculative language in biomedical texts using Recurrent Neural Tensor Networks. RNTN achieves F1=0.885 vs SVM baseline F1=0.881. Paragraph Vector underperforms (F1=0.368). Applications: information retrieval, multi-document summarization.

Summary generated by Claude — human-verified