Thank you for your great work!
Is it possible to provide us the original keyphrases of SemEval2010's test dataset without stemming? Like these files included in the training dataset: "train.combined.final", "train.author.final", and "train.reader.final"