DuRecDial 2.0
收藏资源简介:
DuRecDial 2.0是由百度公司创建的双语平行对话推荐数据集,包含16500条英汉对话,总计255000条语句。该数据集通过严格的质量控制程序由众包工作者标注,覆盖电影、音乐、明星、食品、餐厅、新闻、天气等多个领域。数据集的创建旨在推动多语言和跨语言对话推荐的研究,通过定义五种任务,包括单语、多语和跨语言对话推荐,为模型性能评估提供了一个挑战性的测试平台。
DuRecDial 2.0 is a bilingual parallel conversational recommendation dataset developed by Baidu. It contains 16,500 English-Chinese dialogues, totaling 255,000 utterances. Annotated by crowdworkers through strict quality control procedures, the dataset covers multiple domains such as movies, music, celebrities, food, restaurants, news, and weather. Created to advance research on multilingual and cross-lingual conversational recommendation, it provides a challenging testbed for model performance evaluation by defining five tasks, including monolingual, multilingual, and cross-lingual conversational recommendation.




