Bi-Attention HateXplain : Taking into account the sequential aspect of data during explainability in a multi-task context
This work addresses the need for more reliable and transparent hate speech detection systems, which is crucial for mitigating online threats, though it is incremental as it builds on existing multi-task approaches.
The paper tackles the problem of inconsistent attention and unstable predictions in multi-task hate speech detection models by proposing a bidirectional attention BiRNN model that incorporates sequential data aspects, resulting in improved detection performance, explainability, and reduced unintentional bias on the HateXplain benchmark.
Technological advances in the Internet and online social networks have brought many benefits to humanity. At the same time, this growth has led to an increase in hate speech, the main global threat. To improve the reliability of black-box models used for hate speech detection, post-hoc approaches such as LIME, SHAP, and LRP provide the explanation after training the classification model. In contrast, multi-task approaches based on the HateXplain benchmark learn to explain and classify simultaneously. However, results from HateXplain-based algorithms show that predicted attention varies considerably when it should be constant. This attention variability can lead to inconsistent interpretations, instability of predictions, and learning difficulties. To solve this problem, we propose the BiAtt-BiRNN-HateXplain (Bidirectional Attention BiRNN HateXplain) model which is easier to explain compared to LLMs which are more complex in view of the need for transparency, and will take into account the sequential aspect of the input data during explainability thanks to a BiRNN layer. Thus, if the explanation is correctly estimated, thanks to multi-task learning (explainability and classification task), the model could classify better and commit fewer unintentional bias errors related to communities. The experimental results on HateXplain data show a clear improvement in detection performance, explainability and a reduction in unintentional bias.