COSTRA 1.0

COSTRA 1.0 是一个复杂句子转换的数据集。该数据集旨在研究句子级嵌入，而不仅仅是简单的单词替换或标准释义。数据集的第一个版本仅限于捷克语的句子，但构造方法是通用的，作者计划将其也用于其他语言。该数据集由 4,262 个独特的句子组成，平均长度为 10 个单词，说明了 15 种类型的修改，例如简化、泛化或正式和非正式的语言变体。

COSTRA 1.0 is a dataset for complex sentence transformation. This dataset is designed for research on sentence-level embeddings, rather than solely focusing on simple word replacement or standard paraphrasing. The first version of this dataset is limited to Czech sentences, but its construction methodology is generalizable, and the authors plan to apply it to other languages as well. This dataset consists of 4,262 unique sentences with an average length of 10 words, covering 15 types of modifications such as simplification, generalization, and formal and informal language variants.

搜集汇总

背景与挑战

背景概述

COSTRA 1.0是一个专注于复杂句子转换的数据集，旨在研究句子级嵌入，而非简单单词替换或标准释义。该数据集基于捷克语，包含4,262个独特句子，平均长度为10个单词，并涵盖了15种修改类型，例如简化、泛化以及正式和非正式的语言变体。