Given a gold standard of events and/or times and a system response, we can now get the precision and recall numbers. What we do not get is an easy way to analyze individual errors. Create a little tool that displays mismatches and shows some of the context (for example, showing chunk and sentence boundaries).
Given a gold standard of events and/or times and a system response, we can now get the precision and recall numbers. What we do not get is an easy way to analyze individual errors. Create a little tool that displays mismatches and shows some of the context (for example, showing chunk and sentence boundaries).