Text Expansion (TE) 数据集

Text Expansion (TE) 数据集是由腾讯AI实验室和鹏城实验室共同创建，旨在通过插入细粒度修饰语来丰富或活跃人类写作。数据集包含1200万自动生成的实例和2000个人工标注的参考文献，适用于中英文。创建过程中，采用了多种互补方法确保数据的全面性和多样性。该数据集主要用于智能写作助手中的文本扩展任务，旨在提高文本的具体性和生动性，同时保持基本语义不变。

Text Expansion (TE) dataset was co-developed by Tencent AI Lab and Peng Cheng Laboratory, aiming to enrich or enliven human writing by inserting fine-grained modifiers. It contains 12 million automatically generated instances and 2,000 manually annotated reference samples, and is applicable to both Chinese and English. Multiple complementary methods were adopted during its construction to ensure the comprehensiveness and diversity of the dataset. This dataset is primarily used for text expansion tasks in intelligent writing assistants, with the goal of enhancing the specificity and vividness of written content while retaining the original core semantic meaning unchanged.

搜集汇总

背景与挑战

背景概述

Text Expansion (TE) 数据集是由腾讯AI实验室和鹏城实验室共同创建，用于智能写作助手中的文本扩展任务。它包含1200万自动生成的实例和2000个人工标注的参考文献，适用于中英文，旨在通过插入细粒度修饰语来丰富文本，提高具体性和生动性，同时保持基本语义不变。