AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Style-transfer and Paraphrase: Looking for a Sensible Semantic Similarity Metric

Ivan P. Yamshchikov, Viacheslav Shibaev, Nikolay Khlebnikov, Alexey Tikhonov

AAAI 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 31 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use WMD rather than the other evaluated metrics to estimate semantic similarityParaphrase and style-transfer evaluation · 14 datasetscorrelation with human evaluation: 0.92Improved
Use Word2Vec cosine similarity as a semantic-preservation metricParaphrase and style-transfer evaluation · 14 datasetscorrelation with human evaluation: 0.46Hurt
Use FastText cosine similarity as a semantic-preservation metricParaphrase and style-transfer evaluation · 14 datasetscorrelation with human evaluation: 0.52Hurt
Use BLEU for quantitative semantic similarity estimates rather than binary separation of random pairsParaphrase and style-transfer evaluation · Random sentence pairs across datasetsvariability on random sentences: 0.2%Mixed
Treat sentiment transfer as ordinary style transfer while assuming sentiment is not semanticHuman assessment of Yelp! rewrites · Yelp! rewritehuman labeling: 2.68 ± 0.83Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy