Hey,
In the paper you mention that when evaluating the model on NUS dataset you union the keyphrases assigned by authors and annotators. Do you also union the controlled and uncontrolled keyphrases from Inspec dataset?
Hi,
For the Inspec Dataset, we followed TopicRank paper (and original Hulth's paper (2003) ) and only used the uncontrolled keyphrases (and apply stemming).
Hey, In the paper you mention that when evaluating the model on NUS dataset you union the keyphrases assigned by authors and annotators. Do you also union the controlled and uncontrolled keyphrases from Inspec dataset?