NLI Diversity is limited by the scope of the NLI model's training data

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• What is the effect of more generalized NLI training data on NLI Diversity performance?question0.857
  Future work question arising from the finding that Combined model did not outperform MNLI model
• Confidence NLI Diversity achieves state-of-the-art performance on measuring semantic diversityclaim0.834
  Main performance claim of the paper
• NLI Diversitymethod0.819
  Novel metric proposed in this paper using NLI predictions to score semantic diversity of a response set
• Baseline NLI Diversitymethod0.815
  First variant: aggregates argmax NLI class predictions with contradiction=+1, entailment=-1, neutral=0
• What is the role of neutral NLI predictions in measuring semantic diversity?question0.801
  Motivates the creation of Neutral NLI Diversity as an ablation
• Confidence NLI Diversitymethod0.797
  Best-performing variant: aggregates softmax probability mass rather than binary class counts
• Neutral NLI predictions may capture lexical rather than semantic diversityhypothesis0.797
  Hypothesis proposed to explain Neutral NLI Diversity's high performance on decTest but low on conTest
• Neutral NLI Diversitymethod0.796
  Variant weighting neutral predictions equally to contradictions to test if neutrals capture lexical diversity