TalkDown: A Corpus for Condescension Detection in Context
This addresses the challenge of condescension detection for online communities, which can help mitigate toxic interactions, but it is incremental as it builds on existing methods with a new dataset.
The paper tackles the problem of detecting condescending language in dialogues by introducing TalkDown, a labeled dataset that incorporates discourse context, showing that extending a language-only model with discourse representations improves performance, with specific gains in metrics like F1 scores (e.g., from 0.65 to 0.72).
Condescending language use is caustic; it can bring dialogues to an end and bifurcate communities. Thus, systems for condescension detection could have a large positive impact. A challenge here is that condescension is often impossible to detect from isolated utterances, as it depends on the discourse and social context. To address this, we present TalkDown, a new labeled dataset of condescending linguistic acts in context. We show that extending a language-only model with representations of the discourse improves performance, and we motivate techniques for dealing with the low rates of condescension overall. We also use our model to estimate condescension rates in various online communities and relate these differences to differing community norms.