For classification tasks, submissions will be evaluated and ranked in terms of F1 measure, whereas for the regression task, the root mean squared error will be employed.

Participants are allowed to make two submissions as two csv files where each row corresponds to a tweet. The csv file must be formatted as follows:

Whenever the participants aim the task 1 (prejudical humour recognition task) the csv file must contain humour column with values from {0, 1}, where 0 corresponds to messages whose author is not intended to cause humour and 1 in the opposite situation.

Columns prejudice_group1, prejudice_group2, prejudice_group3 and prejudice_group4 with domain {0, 1},  correspond to the multilabel classification task 2A. Here, group1 refers to women and feminists, group2 to the LGBTIQ community, group3 to immigrants and racially discriminated people and group4 to overweight people.

For participating in task 2B the csv file must contain the prejudice_degree column with a real value from 1 to 5.

*For each submission, please send an email to rlabtam@posgrado.upv.es containing an attached file with your predictions in CSV format. Please name the file using the following format: TEAM_SubmissionID.csv, where TEAM is replaced with your team name and submissionID with an identifier for your submission.