知识卡片
潜在语义分析主题未必可命名
LSA从词项共现的低秩结构得到坐标,某一维可同时涉及人认为不同甚至相反的词。人为命名可帮助沟通,但不能把名字当模型实际理解或毒性证据;不命名也可用这些特征做检索和分类。
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
- · 从词数到语义分析
原书出处 · 1 处
- · 查看原书依据
These new topics will not necessarily correlate to what we humans think about as topics, like pets or history. The machine doesn’t understand what combinations of words mean, just that they go together. When it sees words like dog, cat, and love together a lot, it puts them together in a topic but doesn’t know that such a topic is likely about pets. It might
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第04章《从词数到语义分析》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-4.html)
- 结论依据:摘录中关于潜在语义分析主题未必可命名的定义、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:These new topics will not necessarily correlate to what we humans think about as topics, like pets or history. The machine doesn’t understand what combinations of words mean, just that they go together. When it sees words like dog, cat, and love together a lot, it puts them together in a topic but doesn’t know that such a topic is likely about pets. It might