Banking Domain Dialogue Dataset
收藏arXiv2019-10-16 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/1910.07357v1
下载链接
链接失效反馈官方服务：
资源简介：
本数据集名为Banking Domain Dialogue Dataset，由纽约州立大学布法罗分校和Fondazione Bruno Kessler合作创建。数据集包含200条对话，专注于银行业务领域，如转账、查询余额等。创建过程采用对话自玩策略，模拟用户与对话代理之间的任务导向对话。该数据集旨在通过合成挑战数据，评估神经网络模型在处理未见对话模式时的性能，特别适用于研究神经对话代理的解释性和学习能力。
This dataset, termed Banking Domain Dialogue Dataset, was jointly developed by University at Buffalo, State University of New York and Fondazione Bruno Kessler. It contains 200 dialogues focusing on the banking business domain, such as fund transfer, balance inquiry and other similar scenarios. The dataset was constructed using the self-play dialogue strategy, which simulates task-oriented dialogues between users and dialogue agents. This dataset aims to evaluate the performance of neural network models when handling unseen dialogue patterns by providing synthetic challenge data, and is particularly suitable for researching the interpretability and learning capabilities of neural dialogue agents.
提供机构：
纽约州立大学布法罗分校
创建时间：
2019-10-16
