I had a doubt regarding making submissions for task 2 in the English dataset. From the test data provided, are we supposed to run all tweets through the classifier built for task 1, and then predict on the tweets predicted as Hate from the first classifier to split them into the 3 further classes, or can we filter out the tweets that have None as an ID in the task 2 column and then run inference on only the remaining tweets?Posted by: anand_s @ Sept. 19, 2020, 9:47 a.m.
Thanks for your question.
The prediction file should need to comply with the number of rows (and columns) given in the test set. So in inference stage, you need to perform it for all of the tweets.
Enjoy the competition.
HASOC Competition Organization Team
I had a question regarding the total number of submissions that we are allowed to make. Given that the deadline has been extended, are the total number of submissions still capped at 5 submissions or will that be increased?Posted by: tathagataraha @ Sept. 19, 2020, 3:02 p.m.
The submission limit for each user follows as usual. However, if your submission found an issue (or mistakenly submitted in different sub-tasks), we will inform the participant and might ask for resubmission in this case.Posted by: amitkumarjaiswal @ Sept. 19, 2020, 7:43 p.m.
Do we have to make a submission for each and every language one by one or should we include all the 6 csv files in one zip file?Posted by: tathagataraha @ Sept. 26, 2020, 8:07 p.m.
Yes, you need to do separate submissions for each of the language subtasks.Posted by: amitkumarjaiswal @ Sept. 26, 2020, 8:30 p.m.