Papers by Björn Ross
Compositional Generalisation for Explainable Hate Speech Detection (2025.emnlp-main)
Copied to clipboard
| Challenge: | Hate speech detection models struggle to generalise beyond their training data . dataset biases and the use of sentence-level labels fail to teach the underlying structure of hate speech. |
| Approach: | They propose to use a dataset to train models with fine-grained, span-level annotations . they find that combinations of expressions that deviate from those seen during training are difficult to detect . |
| Outcome: | The proposed model can generalise to a dataset with equal frequency across all contexts while achieving state-of-the-art performance on the human-sourced PLEAD. |
Validating Automatic Evaluation of Controllable Counterspeech Generation: Rankings Matter More Than Scores (2026.eacl-long)
Copied to clipboard
| Challenge: | Existing methods for evaluating attributes of counterspeech are limited and the validity of such evaluations is questionable when the classifiers themselves have only modest performance. |
| Approach: | They examine the automatic evaluation of counterspeech attributes using a multi-attribute counterseech dataset containing 2,728 samples. |
| Outcome: | The proposed model can be trusted by classifier validation, and it can rank models with confidence. |
Cross-lingual Transfer Can Worsen Bias in Sentiment Analysis (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing evidence of demographic bias in SA systems is limited to a handful of languages, and it is costly to create supervised training data in a new language. |
| Approach: | They use counterfactual evaluation to test whether gender or racial biases are imported when using cross-lingual transfer . r&r is much more prevalent than gender biase . |
| Outcome: | The proposed model is compared with monolingual systems in five languages and shows that it is biased more than monolingual ones. |
Explainable Abuse Detection as Intent Classification and Slot Filling (2022.tacl-1)
Copied to clipboard
| Challenge: | Existing models learn what abuse is from labeled examples and base their predictions on spurious cues. |
| Approach: | They propose a machine-friendly representation of the policy that moderators wish to enforce by breaking it down into intents and slots. |
| Outcome: | The proposed model breaks down a social media policy into intents and slots and provides a rationale for its decisions. |
Explainability and Hate Speech: Structured Explanations Make Social Media Moderators Faster (2024.acl-short)
Copied to clipboard
Agostina Calabrese, Leonardo Neves, Neil Shah, Maarten Bos, Björn Ross, Mirella Lapata, Francesco Barbieri
| Challenge: | Existing studies have shown that explanations can support content moderators to make faster decisions, but the benefits of such models have not been studied. |
| Approach: | They propose to use structured explanations to support content moderators to make faster decisions by 7.4%. |
| Outcome: | The proposed models lower the speed of real-world moderators by 7.4% compared to generic explanations and are often ignored . previous studies have shown that explanations can support moderator's decision making by detecting violations of policies but the benefits have not been studied . |