With no explanation, label text_A→text_B with either "strengthener" or "weakener".
text_A: PersonX draws PersonY inference PersonX is seen as good.
text_B: PersonY ignores PersonX
weakener.