Which advanced technique is often employed in abstractive summarization models to ensure that the generated summary remains factually consistent with the source document, mitigating the risk of 'hallucinations'?
Question 2
When evaluating the quality of an abstractive summary, why is human evaluation often considered indispensable, particularly for nuanced aspects like 'fluency' and 'coherence', despite the existence of automated metrics?
Question 3
Consider a scenario where a summarization system is designed to produce a summary of a legal document. The primary requirement is to ensure that all critical legal clauses are present in the summary, even if it means including some redundant information. Which evaluation metric would be most appropriate to prioritize in this context?
Question 4
In the context of multi-sentence summarization, what is the primary purpose of a 'content selection' module in a planning phase?
Question 5
Which of the following best describes the 'redundancy' problem in summarization, particularly when generating multi-sentence summaries?
Summarization Quiz — Natural Language Processing | A-Warded
