Models to extract stance targets (in this case, noun-phrases) from documents expressing stance on those stance targets.