CCPM (Chinese Classical Poetry Matching)

介绍
CCPM是一个大型的中国古典诗歌匹配数据集，可用于诗歌匹配、理解和翻译。
该数据集的主要任务是：给定现代汉语的描述，该模型应该从四个候选中选择与给定描述在语义上最匹配的一行中国古典诗歌。
尺寸
它总共包含 27,218 个实例，分为训练 (21,778)、验证 (2,720) 和测试 (2,720) 集。
格式
每个实例由翻译（现代汉语描述，一个字符串）、选择（中国古典诗歌四个候选行，一个列表）和答案（正确行的索引，0到3之间的整数）组成。

Introduction
CCPM is a large-scale Chinese classical poetry matching dataset designed for poetry matching, comprehension and translation tasks.
The core task of this dataset is: Given a modern Chinese description, the model is required to select the single line of classical Chinese poetry that best matches the given description semantically from four candidate options.

Dataset Size
It contains a total of 27,218 instances, split into training (21,778), validation (2,720) and test (2,720) sets.

Data Format
Each instance consists of three components: a translation (modern Chinese description, a string), a selection (a list of four candidate lines of classical Chinese poetry), and an answer (the index of the correct line, an integer between 0 and 3).

搜集汇总

背景与挑战

背景概述

CCPM是一个大型中国古典诗歌匹配数据集，用于根据现代汉语描述从四个候选诗歌行中选取语义最匹配的一行。该数据集包含27,218个实例，分为训练、验证和测试集，由清华大学于2021年发布。