LLM synthetic generation to enhance online content moderation generalization in hate speech scenarios | Publicación