generalfactneutralMasking-based adjustment representations can preserve overlap for causal text analysisMachine Learning (Statistics)01 Aug 2026http://arxiv.org/abs/2607.26309v1