Robust Neural Abstractive Summarization Systems and Evaluation against Adversarial Information


Oct 14, 2018 Blind Submission readers: everyone
  • Abstract: Sequence-to-sequence (seq2seq) neural models have been actively investigated for abstractive summarization. Nevertheless, existing neural abstractive systems frequently generate factually incorrect summaries and are vulnerable to adversarial information, suggesting a crucial lack of semantic understanding. In this paper, we propose a novel semantic-aware neural abstractive summarization model that learns to generate high quality summaries through semantic interpretation over salient content. A novel evaluation scheme with adversarial samples is introduced to measure how well a model identifies off-topic information, where our model yields significantly better performance than the popular pointer-generator summarizer. Human evaluation also confirms that our system summaries are uniformly more informative and faithful as well as less redundant than the seq2seq model.
  • TL;DR: We propose a semantic-aware neural abstractive summarization model and a novel automatic summarization evaluation scheme that measures how well a model identifies off-topic information from adversarial samples.
  • Keywords: natural language processing, abstractive summarization, adversarial samples
0 Replies