Such papers would probably make it explicit that a given methodology was inadequate (and explain why) so while it may be challenging, it's still a solvable NLP problem.