Topics of interest include, but are not limited to:
Datasets, and Benchmarks for Equality, Diversity, and Inclusion
Construction and annotation of datasets for EDI, including benchmarks for bias detection and mitigation.
Compilation of resources curated for fairness, inclusivity, and accessibility.
Methodologies for annotating intersectional identities (gender, race, disability, religion, sexual orientation, etc.).
Bias Detection and Mitigation in LLMs
Techniques for identifying, measuring, and mediating gender, racial, disability, and other societal biases in NLP and LLMs.
The impact of bias in deployed NLP/LLM systems
Gender-neutral modeling and representational fairness in LLMs
Detection and mitigation of intersectional biases including gender, racial, gender identity, disability, and other societal biases.
Advances in bias mitigation in large language models: in-context learning, prompt engineering, conditional text generation, and adversarial training.
Inclusive Language and Counter-Narratives for LLMs
Algorithms and resources for inclusive language generation with LLMs.
Counter-narrative modeling for combating toxicity, hate speech, and misinformation targeting marginalized communities.
Dialogue systems and multi-agent approaches that align with inclusivity goals.
Human-in-the-loop and participatory strategies for enhancing inclusiveness.
Multilingual and Multicultural Approaches for LLMs
Multicultural and multilingual LLMs and approaches
Speech and language recognition for minority and under-resourced groups
Code-mixed and cross-lingual approaches for inclusive technologies
Responsible, Explainable, and Trustworthy LLMs for EDI
Detecting and mitigating hallucinations, misinformation, and toxicity in LLM systems
Explainable and trustworthy LLMs
Evaluation frameworks incorporating ethics, accountability, and transparency