知识卡片
数据切分要复现冷启动条件
按评分留出可检查已有用户的未知偏好,按用户留出可检查训练中未见用户,进一步隔离目标域则改变可用迁移信息。这些协议对应不同问题。普通随机留出不能直接证明无目标域历史的冷启动效果。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- · 查看原书依据
Hold-out Fig.(9)-center can be adopted whenever the test profiles set is a derivative of the training profiles set, including ratings from source and target domains. Besides that, the test profile set can be sampled and hidden from the initial dataset, assuming all domains and without the users being split. If the main aim is to increase the accuracy, this partitioning type can measure linked- and
参考来源
- 位置:《Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications》第04章《多领域情境推荐的张量分解》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Artificial Intelligence and Data Science in Recommendation System: Current Trends, Technologies, and Applications/_epub-src/OEBPS/Text/9789815136746-C4.html)
- 结论依据:摘录中关于数据切分要复现冷启动条件的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:Hold-out Fig.(9)-center can be adopted whenever the test profiles set is a derivative of the training profiles set, including ratings from source and target domains. Besides that, the test profile set can be sampled and hidden from the initial dataset, assuming all domains and without the users being split. If the main aim is to increase the accuracy, this partitioning type can measure linked- and