Skip to content

Latest commit

 

History

History
13 lines (13 loc) · 906 Bytes

ASFAS.md

File metadata and controls

13 lines (13 loc) · 906 Bytes

Analyzing sentence fusion in Abstractive Summarization

  • The paper focuses on analyzing summaries formed by sentence fusion on the basis of faithfulness, grammaticality and method of fusion for summary sentences.
  • Findings
    • System outputs are more likely to be not factual compared to being ungrammatical.
    • Systems oftens simply concatenate rather than entity replacement while performing sentence fusion.
      • They are also generally more faithful.
    • Systems struggle to reliably perform complex fusion because they lead to incorrect facts.
      • System summary sentences formed by fusion also have low faithfulness compared to reference summaries.
  • Different methods of fusion are
    • Replacement (usually entity/pronoun)
    • Balanced concatenation (Similar length sentence fusion)
    • Imbalanced concatenation (Dissimilar length sentence fusion)
    • All other cases.