Many NLP tasks can be regarded as a selection problem from set of options, such classification tasks, multi-choice question answering, etc. Textual entailment (TE) has been shown the state-of-the-art (SOTA) approach to dealing with those problems. TE treats input texts premises (P), options hypotheses (H), then handles by modeling (P, H) pairwise. Two limitations: first, pairwise is unaware oth...
