How to solve the problem that topK's K is different for every input text?

[guotong1988]

The output topK's K is fixed now.

Do you think training a classifier to predict the value of K for every input is a good solution?

Thank you very much.

[RandolphVI]

@guotong1988

First, how to solve the problem that K of topK is different for every input text?

• You can use some evaluation metrics which are not related to the threshold (e.g., AUPRC).

Second, training a classifier to predict the value of K for every input is a good solution?

• It's an interesting idea that I wanted to do before. There are also some papers that do similar work, you can refer to them.

Hope it helps!

[guotong1988]

Thank you. In summary, threshold is one solution. The solution of mine is ok. Am I right?

[RandolphVI]

@guotong1988 Generally speaking, yes.