Given a set of values, ConflictScope generates scenarios in which an LLM-based assistant faces a conflict between a pair of values in the set. It then evaluates which value a target LLM supports more in each scenario before combining scenario-level judgments into a value ranking.
