Empirical Analysis of Multi-Task Learning for Reducing Model Bias in Toxic Comment Detection
arXiv:1909.09758
Abstract
With the recent rise of toxicity in online conversations on social media platforms, using modern machine learning algorithms for toxic comment detection has become a central focus of many online applications. Researchers and companies have developed a variety of models to identify toxicity in online conversations, reviews, or comments with mixed successes. However, many existing approaches have learned to incorrectly associate non-toxic comments that have certain trigger-words (e.g. gay, lesbian, black, muslim) as a potential source of toxicity. In this paper, we evaluate several state-of-the-art models with the specific focus of reducing model bias towards these commonly-attacked identity groups. We propose a multi-task learning model with an attention layer that jointly learns to predict the toxicity of a comment as well as the identities present in the comments in order to reduce this bias. We then compare our model to an array of shallow and deep-learning models using metrics designed especially to test for unintended model bias within these identity groups.
arXiv commentICWSM 2020
References in corpus (12)
- Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling
- Semantics derived automatically from language corpora contain human-like biases
- An Overview of Multi-Task Learning in Deep Neural Networks
- Word Embeddings Quantify 100 Years of Gender and Ethnic Stereotypes
- The Measure and Mismeasure of Fairness
- Automated Hate Speech Detection and the Problem of Offensive Language
- Feed-Forward Networks with Attention Can Solve Some Long-Term Memory Problems
- Learning Multiple Tasks with Multilinear Relationship Networks
- Conversations Gone Awry: Detecting Early Signs of Conversational Failure
- Machine Learning Suites for Online Toxicity Detection
- Deep Automated Multi-task Learning
- Stop Illegal Comments: A Multi-Task Deep Learning Approach