no code implementations • 21 Feb 2024 • Preni Golazizian, Ali Omrani, Alireza S. Ziabari, Morteza Dehghani
In subjective NLP tasks, where a single ground truth does not exist, the inclusion of diverse annotators becomes crucial as their unique perspectives significantly influence the annotations.
1 code implementation • 29 Sep 2023 • Ali Omrani, Alireza S. Ziabari, Preni Golazizian, Jeffery Sorensen, Morteza Dehghani
Detecting problematic content, such as hate speech, is a multifaceted and ever-changing task, influenced by social dynamics, user populations, diversity of sources, and evolving language.
no code implementations • 11 Oct 2022 • Ali Omrani, Brendan Kennedy, Mohammad Atari, Morteza Dehghani
Existing word embedding debiasing methods require social-group-specific word pairs (e. g., "man"-"woman") for each social attribute (e. g., gender), which cannot be used to mitigate bias for other social groups, making these methods impractical or costly to incorporate understudied social groups in debiasing.
no code implementations • 10 Aug 2022 • Jackson Trager, Alireza S. Ziabari, Aida Mostafazadeh Davani, Preni Golazizian, Farzan Karimi-Malekabadi, Ali Omrani, Zhihe Li, Brendan Kennedy, Nils Karl Reimer, Melissa Reyes, Kelsey Cheng, Mellow Wei, Christina Merrifield, Arta Khosravi, Evans Alvarez, Morteza Dehghani
Moral framing and sentiment can affect a variety of online and offline behaviors, including donation, pro-environmental action, political engagement, and even participation in violent protests.
no code implementations • ACL (WOAH) 2021 • Aida Mostafazadeh Davani, Ali Omrani, Brendan Kennedy, Mohammad Atari, Xiang Ren, Morteza Dehghani
By applying logit pairing to equalize outcomes on the restricted set of counterfactuals for each instance, we improve fairness metrics while preserving model performance on hate speech detection.
no code implementations • 24 Oct 2020 • Aida Mostafazadeh Davani, Ali Omrani, Brendan Kennedy, Mohammad Atari, Xiang Ren, Morteza Dehghani
Counterfactual token fairness for a mentioned social group evaluates the model's predictions as to whether they are the same for (a) the actual sentence and (b) a counterfactual instance, which is generated by changing the mentioned social group in the sentence.