Dear participants,
Here is one clarification about the data for task 3: As you may have noticed, not all the "es" tags have the corresponding BIO information for task 3. This is because, given the complexity of the task and the time we had, we could only annotate it for half of the examples. However, if you want to participate in task 3, your systems should always output a BIO tag for all the "es" labels (es-b-ul, es-i-ul, es-b-cc, es-i-cc). The evaluation script will only calculate the metric restricted to the test examples that have the task 3 information, and ignore it for the rest.
Regards,
Luis