Jonathan Oheix

I just figured out that I was wrong in my calculation of the NDCG in the previous...

Maybe there is something I did wrong concerning the origin of my data ? Here's how...

I tried to train the model and test on held-out data but I obtained results that...

I tried to train the model and test on held-out data but I obtained results that...

Ok I removed the "qid:" part. But now I have the same too optimistic results as previously......

Normally not, because my qrel file corresponds to the training set with only the...

I don't have a separate test set because I use a k-fold cross validation to compute...

I don't have a separate test set because I use a k-fold cross validation to compute...