The query topic (the training set) is defined by:

The abstracts to be ranked (the test set) are defined by: