Competition

Evaluation Criteria

Classification systems will be evaluated using the macro-averaged F1-score.

The unlabeled test set can be obtained here

Submission format

The prediction file format is a simple comma separated file, where each line consist of two fields: the document id and the label (either OFF or NOT). The ID field shoud match the ID's of the documents in the test file. The submitted CSV file should have exactly the same number (3528) of instances as the test file, but it shoud not have a haeder line. Please also include a README.txt file that briefly describes the approach used in the submission. Both files should be packaged together as a zip file and submitted through CodaLab. The file starting kit contains an example submission with a majority class baseline (all labels are OFF).

Terms and Conditions

The data in this competition, OffensEval 2020 - Turkish, is licensed under a Creative Commons Attribution licence (CC-BY). If you use this data, please acknowledge the following work:

A Corpus of Turkish Offensive Language on Social Media, Çağrı Çöltekin (2020), Proceedings of LREC

@inproceedings{sigurbergsson2020offensive,
    title={A Corpus of Turkish Offensive Language on Social Media},
    author={\c{C}\"{o}ltekin, \c{C}a\u{g}r{\i}},
    year={2020},
    inproceedings={Proceedings of the 12th International Conference on Language Resources and Evaluation},
    organization={ELRA}
}