ArgMining22-Shared-Task

Shared task predicting validity and novelty

View the Project on GitHub phhei/ArgsValidNovel

How to submit

In order to keep it simple, we ask you to write an e-mail to pheinisch@techfak.uni-bielefeld.de with opitz@cl.uni-heidelberg.de in Cc. We ask to contribute exactly one approach per e-mail. If you have developed several approaches/ systems, write multiple e-mails, one e-mail for one approach/ system.

We remember our terms:

There are up to 5 submissions from different approaches (systems) allowed per team and per subtask. The submission must be submitted via e-mail. You are allowed to withdraw your submission at anytime until the final deadline (also via e-mail).

You are free to participate only in one subtask or both.

Please read the following e-mail-instructions carefully since we can only accept e-mails following the given instructions:

How to structure a submission e-mail

You have to structure your e-mail in the following way:

Subject of the e-mail

[ArgMining22-SharedTask-Subtask (A or B) ] team name (number of appraoch)

(the number of approach is an incrementing number starting with 1 to identify this submission. If you just submit one approach/ submission (per subtask), it’s just [ArgMining22-SharedTask-Subtask (A or B) ] team name (1). However, let’s say you submit your already third approach for subtask A, the subject should be [ArgMining22-SharedTask-Subtask A] team name (3).)

Text body of the e-mail

1. General information

Please provide a list of information containing the following information in the following order:

  1. The name of your team (if you submit without having a team, you can write your name here)
  2. All team members’ full names and e-mail addresses, separated by a comma. The first name in the list should be the main contact person
  3. (optional: the affiliation of each team member)
  4. the name/ short title of your approach

2. Abstract describing your approach/ system

Please describe in 5-10 sentences your approach/ system.

3. Additional [training] data used?

If you used additional [training] data (data differing from the given training data and/or evaluation data), please descibe here:

Appendix

Only 1 single csv file (please ensure an uft-8-encoding)! Don’t use an archive file (hence, no .zip files)!

The csv file should contain the predictions for the test data and should be structured as follows:

In the case of Subtask A

topic, Premise, Conclusion, predicted validity, predicted novelty

The predicted validity/ novelty has to be either -1 (not valid/ novel) or 1 (valid/ novel). No floating numbers!

In the case of Subtask B

topic, Premise, Conclusion 1, Conclusion 2, predicted validity, predicted novelty

The predicted validity/ novelty has to be either -1 (conclusion 1 outperforms) or 0 (tie) or 1 (conclusion 2 outperforms). No floating numbers!