I am summarizing an research article who use PU learning.
The goal of the research is to predict the probability of whether a startup can secure round funding.
This is a binary classification model.
They detect their data directly from open, public websites. In the websites, they can find the information about a startup have raised a fund. They label those as positive.
They did not consider those startup who cannot find secure found funding information online as negative, instead, consider them as unlabeled to prevent the noise from undetected positive.
Does my logic for explain a positive unlabeled learning correct?