I'd like to remind you all that your challenge test predictions are
due in just over a week, on April 13. Please email me your test
predictions at or before then. In the following week we'll make the
labelled test data available to everyone so that you can perform your
own error analysis.
Also when you submit your system predictions, can you please include a
paragraph or two describing your system? Include an outline of your
algorithm, its data requirements and any novel features of your
approach. This will give us something to work with when writing the
overall results paper.
Finally, I've added some new tools to the shared task homepage -
namely code for evaluation (please use this to road-test your models'
predictions on the development set) and various baselines (brown word
clustering, DMV). Please tell me if you have any problems with this.
Best,
Trevor
> I was getting errors in the python script that removed the punctuation
> (something about the printing of the file). Then I looked at the code and
> saw that in line 58 you were using "{}\t" to print the different fields.
> I'm not sure what exactly that notation is supposed to do, but I changed all
> the {} to %s and the script work.
This is a python 2.7 feature, sorry we should have stated that this is required
(or just make the change you suggest to %s format specifiers). This is used in
quite a few places in the code, so I suggest just upgrading python.
> Also, but unrelated, the link to the evaluation scripts is broken.
Oops, thanks! I've fixed it now.
Best,
Trevor