"Nice Try, Kiddo": Investigating Ad Hominems in Dialogue Responses
arXiv:2010.12820
Abstract
Ad hominem attacks are those that target some feature of a person's character instead of the position the person is maintaining. These attacks are harmful because they propagate implicit biases and diminish a person's credibility. Since dialogue systems respond directly to user input, it is important to study ad hominems in dialogue responses. To this end, we propose categories of ad hominems, compose an annotated dataset, and build a classifier to analyze human and dialogue system responses to English Twitter posts. We specifically compare responses to Twitter topics about marginalized communities (#BlackLivesMatter, #MeToo) versus other topics (#Vegan, #WFH), because the abusive language of ad hominems could further amplify the skew of power away from marginalized populations. Furthermore, we propose a constrained decoding technique that uses salient -gram similarity as a soft constraint for top- sampling to reduce the amount of ad hominems generated. Our results indicate that 1) responses from both humans and DialoGPT contain more ad hominems for discussions around marginalized communities, 2) different quantities of ad hominems in the training data can influence the likelihood of generating ad hominems, and 3) we can use constrained decoding techniques to reduce ad hominems in generated dialogue responses.
arXiv commentNAACL 2021
References in corpus (9)
- The Curious Case of Neural Text Degeneration
- Hierarchical Neural Story Generation
- Plug and Play Language Models: A Simple Approach to Controlled Text Generation
- Before Name-calling: Dynamics and Triggers of Ad Hominem Fallacies in Web Argumentation
- Lexically Constrained Decoding for Sequence Generation Using Grid Beam Search
- Towards Controllable Biases in Language Generation
- Constrained Decoding for Neural NLG from Compositional Representations in Task-Oriented Dialogue
- Detecting Offensive Content in Open-domain Conversations using Two Stage Semi-supervision
- TORQUE: A Reading Comprehension Dataset of Temporal Ordering Questions