Papers with bias identification
CBBQ: A Chinese Bias Benchmark Dataset Curated with Human-AI Collaboration for Large Language Models (2024.lrec-main)
Copied to clipboard
| Challenge: | a dataset of Chinese large language models is used to measure societal biases . many studies have shown that LLMs exhibit harmful societal biased outputs despite human data . |
| Approach: | They present a Chinese Bias Benchmark dataset that includes over 100K questions constructed by human experts and generative language models. |
| Outcome: | The proposed dataset covers stereotypes and societal biases in 14 social dimensions related to Chinese culture and values. |
Hollywood Identity Bias Dataset: A Context Oriented Bias Analysis of Movie Dialogues (2022.lrec-1)
Copied to clipboard
Sandhya Singh, Prapti Roy, Nihar Sahoo, Niteesh Mallela, Himanshu Gupta, Pushpak Bhattacharyya, Milind Savagaonkar, Nidhi Sultan, Roshni Ramnani, Anutosh Maitra, Shubhashis Sengupta
| Challenge: | Movies reflect society and also hold power to transform opinions. |
| Approach: | They propose to annotate movie scripts for identity bias using a dataset that is annotated for gender, race/ethnicity, religion, age, occupation, LGBTQ, and other . |
| Outcome: | The proposed dataset contains dialogue turns annotated for gender, race/ethnicity, religion, age, occupation, LGBTQ, and other, which contains biases like body shaming, personality bias, etc. |
Improving Bias Mitigation through Bias Experts in Natural Language Understanding (2023.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to mitigate the detrimental effect of bias on the network include debiasing methods that down-weight the biased examples identified by an auxiliary model, which is trained with explicit bias labels. |
| Approach: | They propose a framework that introduces binary classifiers between the auxiliary model and main model, coined bias experts, to reduce the detrimental effect of bias on the network. |
| Outcome: | The proposed approach outperforms the state-of-the-art on various datasets while achieving high performance on in-distribution data. |