Code-mixed Natural Language Inference (NLI) Dataset
收藏arXiv2020-04-13 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2004.05051v2
下载链接
链接失效反馈官方服务：
资源简介：
本数据集由微软研究印度创建，专注于代码混合的自然语言推理任务，特别是混合了印地语和英语的对话。数据集包含400个基于宝莱坞电影对话的代码混合前提和2240个由印地语-英语双语者提供的代码混合假设。数据集的创建过程涉及初步注释研究和最终注释协议的制定。该数据集旨在帮助机器理解代码混合语言，特别是在对话代理和聊天机器人中的应用，以解决多语言社区中的沟通问题。
This dataset was developed by Microsoft Research India, targeting code-mixed natural language inference tasks, particularly Hindi-English mixed dialogues. It includes 400 code-mixed premises sourced from dialogues in Bollywood films, and 2240 code-mixed hypotheses provided by Hindi-English bilingual speakers. The dataset construction process involved preliminary annotation research and the establishment of a finalized annotation protocol. This dataset is intended to assist machines in comprehending code-mixed languages, especially for applications in conversational agents and chatbots, to address communication issues in multilingual communities.
提供机构：
微软研究印度
创建时间：
2020-04-10
