gsarti/magpie
收藏资源简介：
MAGPIE语料库（Haagsma et al. 2020）是一个大规模的意义标注语料库，包含潜在的习语表达（PIEs），基于英国国家语料库（BNC）。潜在的习语表达类似于习语表达，但也包括习语表达的字面用法，例如“我在一天结束时下班”中的“一天结束时”。该数据集版本反映了Dankers等人（2022）在研究中使用的过滤子集，用于研究NMT模型如何表示PIEs。作者使用了37k个标注为完全比喻或字面的样本，涉及1482个包含名词、数字或颜色形容词的习语（称为关键词）。由于习语在句法和形态上具有可变性，研究主要集中在名词上。PIEs及其上下文使用原始语料库的词级注释进行分离。
The MAGPIE Corpus (Haagsma et al. 2020) is a large-scale meaning-annotated corpus containing Potential Idiomatic Expressions (PIEs), which was constructed based on the British National Corpus (BNC). Potential Idiomatic Expressions (PIEs) are similar to standard idiomatic expressions, but also cover their literal usage; for instance, the phrase "at the end of the day" in the sentence "I finish work at the end of the day". This specific version of the dataset corresponds to the filtered subset used in the study conducted by Dankers et al. (2022), which investigates how Neural Machine Translation (NMT) models represent PIEs. The authors employed 37,000 samples annotated as either fully figurative or literal, which are associated with 1,482 idioms containing nouns, numbers, or color adjectives—these idioms are referred to as "keywords". Owing to the syntactic and morphological variability inherent in idioms, the study primarily focuses on nouns. PIEs and their surrounding contextual instances were isolated using the word-level annotations from the original corpus.
数据集概述
数据集名称： MAGPIE
数据集描述： MAGPIE是一个大规模的潜在习语表达（PIEs）的感官注释语料库，基于英国国家语料库（BNC）。该数据集包含37,000个样本，注释为完全比喻或字面意义，涉及1482个包含颜色名词、数字或形容词的习语。
语言： 英语（BCP-47 en）
数据集结构：
- 数据实例： 每个实例包含句子、注释、习语、使用方式、变体和词性标签。
- 数据分割： 训练集包含44,451个实例。
许可证： 知识共享4.0许可证（CC-BY-4.0）
任务类别： 文本分类、文本到文本生成、翻译
数据集创建： 由专家生成，参考原始文章MAGPIE: A Large Corpus of Potentially Idiomatic Expressions和Can Transformer be Too Compositional? Analysing Idiom Processing in Neural Machine Translation。
引用信息： bibtex @inproceedings{haagsma-etal-2020-magpie, title = "{MAGPIE}: A Large Corpus of Potentially Idiomatic Expressions", author = "Haagsma, Hessel and Bos, Johan and Nissim, Malvina", booktitle = "Proceedings of the 12th Language Resources and Evaluation Conference", month = may, year = "2020", address = "Marseille, France", publisher = "European Language Resources Association", url = "https://aclanthology.org/2020.lrec-1.35", pages = "279--287", language = "English", ISBN = "979-10-95546-34-4", } @inproceedings{dankers-etal-2022-transformer, title = "Can Transformer be Too Compositional? Analysing Idiom Processing in Neural Machine Translation", author = "Dankers, Verna and Lucas, Christopher and Titov, Ivan", booktitle = "Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)", month = may, year = "2022", address = "Dublin, Ireland", publisher = "Association for Computational Linguistics", url = "https://aclanthology.org/2022.acl-long.252", doi = "10.18653/v1/2022.acl-long.252", pages = "3608--3626", }
