$\texttt{LifeTox}$: Unveiling Implicit Toxicity in $\texttt{Life}$ Advice

Anonymous

TL;DR: LifeTox, the novel dataset designed to unveil implicit toxicity in advice behind various real-life advice-seeking contexts, thereby facilitating the training of robust and generalizable toxicity detectors.

Abstract: As large language models become increasingly integrated into daily life, detecting implicit toxicity across diverse contexts is crucial. To this end, we introduce \texttt{LifeTox}, a dataset designed for identifying implicit toxicity within a broad range of advice-seeking scenarios. Unlike existing safety datasets, \texttt{LifeTox} comprises diverse contexts derived from personal experiences through open-ended questions. Our experiments demonstrate that RoBERTa fine-tuned on \texttt{LifeTox} matches or surpasses the zero-shot performance of large language models in toxicity classification tasks. These results underscore the efficacy of \texttt{LifeTox} in addressing the complex challenges inherent in implicit toxicity.

Paper Type: short

Research Area: Resources and Evaluation

Contribution Types: NLP engineering experiment, Data resources, Data analysis

Languages Studied: English