Large-scale, diverse, paraphrastic bitexts via sampling and clustering
| Autor: | Matt Post, Nils Holzenberger, Benjamin Van Durme, Abhinav Singh, J. Edward Hu |
|---|---|
| Předmět: |
Computer science
business.industry Inference Sampling (statistics) 02 engineering and technology computer.software_genre Paraphrase Task (project management) Resource (project management) 020204 information systems 0202 electrical engineering electronic engineering information engineering Beam search 020201 artificial intelligence & image processing Artificial intelligence business Cluster analysis computer Sentence Natural language processing |
| Zdroj: | Scopus-Elsevier CoNLL |
| Popis: | Producing diverse paraphrases of a sentence is a challenging task. Natural paraphrase corpora are scarce and limited, while existing large-scale resources are automatically generated via back-translation and rely on beam search, which tends to lack diversity. We describe ParaBank 2, a new resource that contains multiple diverse sentential paraphrases, produced from a bilingual corpus using negative constraints, inference sampling, and clustering.We show that ParaBank 2 significantly surpasses prior work in both lexical and syntactic diversity while being meaning-preserving, as measured by human judgments and standardized metrics. Further, we illustrate how such paraphrastic resources may be used to refine contextualized encoders, leading to improvements in downstream tasks. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|