Evaluating Perspectival Biases in Cross-Modal Retrieval
This work introduces the Cross-Cultural, Cross-Modal, Cross-lingual Multimodal (3XCM) benchmark, and suggests that achieving equitable multimodal retrieval necessitates targeted strategies that explicitly decouple language from culture, rather than relying solely on broader data exposure.
Teerapol Saengsukhiran, Peerawat Chomphooyod, Narabodee Rodjananant et al.
· Annual Meeting of the Associ... · 0 citations