The Fact-based Event-centric Human-value Understanding (FEHU) task aims to evaluate language models’ ability to identify human values in factual news articles. It includes two core subtasks.
Task 1, Article-level Human Value Recognition, involves multi-label recognition of human values expressed across entire articles.

Task 2, Subevent-level Human Value Recognition, involves multi-label recognition of human values expressed across every subevent of articles.
  All values are annotated using a structured taxonomy, including labels such as “equality,” “justice,” and “freedom of thought.”

FEHU offers a unique benchmark for evaluating language models’ value alignment and ethical sensitivity in both classification and generation tasks.