: Vocabulario de PLN: ROUGE

NLP resources and evaluation > measure > metric > automatic evaluation metric > ROUGE

ROUGE

A set of metrics commonly used to evaluate text summarization tasks by comparing machine-generated summaries to reference summaries provided by humans. (Based on Santhosh, Understanding BLEU and ROUGE score for NLP evaluation, on medium.com, 2023)

In prior baselines (rows 3 and 5) we observe a drop in ROUGE score with ranges 2.05 -2.14 and 0.06 in terms of BS when switching from oracle to predicted markers. (Elaraby, Zhong & Litman, 2023)
The quality of generated summaries is typically assessed by comparing the generated summary against some reference summary using ROUGE (Recall-Oriented Understudy for Gisting Evaluation) (ROUGE 2004). (de Andrade & Becker, 2023)
These ROUGE scores measure the accuracy based on unigrams bigrams and longest subsequences. (Liu, Sun & Gao, 2021)
To summarize ROUGE scores assess the similarity between candidates and references based on the overlap of unigrams bigrams and the longest common sequence likewise for BLEU; while BLEU focuses on precision ROUGE focuses on recall. (Zhou, Ringeval & Portet, 2023)
To train the regression model we first calculated the ROUGE score for each sentence in the training set regarding the title as a reference summary. (Ishigaki, Takamura & Okumura, 2017)

http://data.loterre.fr/ark:/67375/8LP-GKBJ37VQ-K

RDF/XML TURTLE JSON-LD última modificación 10/10/24

Loterre