Orfeas Menis Mastromichalakis

Machine translation benchmarks have functionally saturated, but pairing human-authored failure cases with deterministic verification rules reveals critical multimodal blind spots that automated metrics consistently miss.

Explanations for hate speech detection can be significantly improved by combining LLMs with curated vocabularies, offering a path to more transparent and accountable content moderation.