The CISP annotation schema uncovers hypotheses and explanations in full-text scientific journal articles

Increasingly, as full-text scientific papers are becoming available, scientific queries have shifted from looking for facts to looking for arguments. Researchers want to know when their colleagues are proposing theories, outlining evidentiary relations, or explaining discrepancies. We show here that sentence-level annotation with the CISP schema adapts well to a corpus of biomedical articles, and we present preliminary results arguing that the CISP schema is uniquely suited to recovering common types of scientific arguments about hypotheses, explanations, and evidence.