KeyAware News

KeyAware News是由四川大学计算机学院和微软亚洲研究院合作创建的大型新闻标题数据集，包含超过18万条新闻文章、标题和关键词的匹配三元组。该数据集通过分析用户搜索查询和点击行为，挖掘出用户感兴趣的关键词，并将其与新闻文章和标题对齐。数据集的创建旨在支持新闻多标题生成模型的训练，解决传统单标题生成方法的局限性，提高新闻标题的多样性和相关性。该数据集的应用领域主要集中在新闻自动化摘要和标题生成，以提升用户体验和新闻编辑效率。

KeyAware News is a large-scale news headline dataset co-developed by the College of Computer Science at Sichuan University and Microsoft Research Asia. It contains over 180,000 matching triplets of news articles, headlines and their associated keywords. The dataset is constructed by analyzing user search queries and click behaviors to extract keywords aligned with user interests, then aligning these keywords with corresponding news articles and their headlines. This dataset is developed to support the training of news multi-headline generation models, addressing the limitations of traditional single-headline generation approaches and improving the diversity and relevance of news headlines. Its primary application scenarios focus on news automatic summarization and headline generation, aiming to enhance user experience and news editing efficiency.