With no explanation, label text_A→text_B with either "strengthener" or "weakener".
text_A: A dog with a black and brown marbled coat scowls at the camera while a young woman in a black suit walks in the background. A tall person in a coat
text_B: The woman ducks under an average sized man.
weakener.