Retrofitting Contextualized Word Embeddings with Paraphrases

@inproceedings{Shi2019RetrofittingCW,
  title={Retrofitting Contextualized Word Embeddings with Paraphrases},
  author={Weijia Shi and Muhao Chen and Pei Zhou and Kai-Wei Chang},
  booktitle={EMNLP},
  year={2019}
}
Contextualized word embeddings, such as ELMo, provide meaningful representations for words and their contexts. They have been shown to have a great impact on downstream applications. However, we observe that the contextualized embeddings of a word might change drastically when its contexts are paraphrased. As these embeddings are over-sensitive to the context, the downstream model may make different predictions when the input sentence is paraphrased. To address this issue, we propose a post… Expand
9 Citations

Figures, Tables, and Topics from this paper

Probing for idiomaticity in vector space models
  • PDF
MULTISEM at SemEval-2020 Task 3: Fine-tuning BERT for Lexical Meaning
  • 1
  • Highly Influenced
  • PDF
Experiments on Paraphrase Identification Using Quora Question Pairs Dataset
  • 2
  • PDF
Cross-lingual Entity Alignment with Incidental Supervision
  • 1
  • PDF
Cross-lingual Entity Alignment for Knowledge Graphs with Incidental Supervision from Free Text
  • 2
  • PDF
Knowledge Acquisition with Transferable and Scalable Representation Learning
  • PDF
Knowledge Acquisition with Transferable and Robust Representation Learning
  • PDF

References

SHOWING 1-10 OF 32 REFERENCES
Deep contextualized word representations
  • 5,630
  • Highly Influential
  • PDF
Learned in Translation: Contextualized Word Vectors
  • 577
  • PDF
Explicit Retrofitting of Distributional Word Vectors
  • 53
  • PDF
Semi-supervised sequence tagging with bidirectional language models
  • 381
  • PDF
Ultradense Word Embeddings by Orthogonal Transformation
  • 94
  • PDF
Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank
  • 4,295
  • PDF
SemEval-2017 Task 1: Semantic Textual Similarity Multilingual and Crosslingual Focused Evaluation
  • 546
  • PDF
Re-examining Machine Translation Metrics for Paraphrase Identification
  • 197
  • PDF
...
1
2
3
4
...