Clickbait Spoiling Task 1 on Spoiler Classification: What should be classified?

To provide participants in the task with as much information as possible, we also publish questions and answers that reach us by mail in this Forum (all participants in the communication agreed).

The following question arrived via mail:

This mail is regarding the Clickbait Spoiling task, which happens to be a part of the Sem Eval workshop 2022-23. My team and I went through the problem statement and the code. We had a question, and it would be really helpful if you could help us with the same.

The code has a separate section about Classification. We’re not completely sure if this is about the binary Classification of a post being clickbait or not or if it is about the multi-class Classification of a clickbait into “phrase”, “passage”, and “multi”.

We provided the following answer:

All posts in our dataset are clickbait posts, and the first task is a multi-class classification: given clickbait post, classify if this clickbait post warrants a “phrase”, “passage”, or “multi” spoiler. While a full solution would be this multi-class classification, we will also evaluate solutions on how well they can distinguish between any two of the classes (which would be helpful in ensemble methods), so we also encourage submissions of classifiers that for instance classify only into “phrase” or “passage”.

Please do not hesitate to contact us in case of further questions!

Best Regards,