> Recall@K

How is the final metric Recall @ K computed?
Is it a mean of (Relevant_Items_Recommended in top-k) / (Relevant_Items) for each query, where Relevant_Items for each test query = 1

Posted by: IvonaTau @ Sept. 24, 2019, 3:24 p.m.

The overall performance is computed by taking the average of Recall@10 and Recall@50 on the three fashion categories

Posted by: yupenggao @ Sept. 24, 2019, 3:29 p.m.

Yes :) my question is how the individual recall@10 and recall@50 are computed ?

E.g for Recall@50 , is it (# of correct guesses present in ranking) / length of test set ?

Posted by: IvonaTau @ Sept. 24, 2019, 3:45 p.m.

Oh, For that, we provided evaluation script in challenge information page. You can take a look. https://codalab.lri.fr/competitions/573#learn_the_details-evaluation

Posted by: yupenggao @ Sept. 24, 2019, 3:58 p.m.
Post in this thread