Business Scene Dialogue

日英商务对话语料库，即商务场景对话语料库，分3步构建：
选择业务场景，
根据所选场景编写单语对话场景，以及
将场景翻译成另一种语言。
一半的单语场景是用日语写的，另一半是用英语写的。整个构建过程由满足以下条件的人员监督，以保证对话自然：
有从事语言学习项目的经验，尤其是商务对话
能够在各种商务场景中用日语和英语流畅地与他人沟通
有涉足商业经验
BSD 语料库分为平衡的训练集、开发集和评估集。这些集合中的文档在场景和原始语言方面是平衡的。在这个存储库中，我们公开共享完整的开发和评估集以及部分训练数据集。

This Japanese-English business dialogue corpus, a scenario-based business conversation corpus, is constructed in three steps:
1. selecting business scenarios;
2. developing monolingual dialogue scenarios based on the selected scenarios;
3. translating the completed scenarios into the other language.

Half of the monolingual scenarios are written in Japanese, while the remaining half are in English. The entire construction process is supervised by personnel meeting the following criteria to ensure the naturalness of the dialogues:
1. Prior experience in language learning programs, especially those related to business dialogue;
2. Proficiency in fluent communication in both Japanese and English across a wide range of business scenarios;
3. Hands-on business experience.

The BSD Corpus is divided into balanced training, development, and evaluation sets. Documents within these sets are balanced in terms of scenario type and source language. In this repository, we publicly share the complete development and evaluation sets, as well as a partial subset of the training dataset.

搜集汇总

背景与挑战

背景概述

Business Scene Dialogue 是一个日英商务对话语料库，通过选择场景、编写对话和翻译构建，并由专业人员监督以确保自然性。该数据集包含平衡的训练、开发和评估集，由东京大学于2019年发布。