Computer scienceData scienceNLPMain NLP tasksText summarization

Overview of text summarization

Limits of automatic evaluation

Report a typo

In the theory section, we mention how automatic evaluation acts as a fast way to estimate the model's quality. But what are the known limitations for some of the widespread metrics for summarization?

Tip: review the Challenges of the automatic summary generation section from the theory.

Select one or more options from the list

Limited correlation with human judgement

Lack of paraphrase detection

Inability to properly capture order

Repetetiveness detection

___

Create a free account to access the full topic