知识卡片
穷举检索复杂度包含维数
N个D维文档逐个与查询点积,直接计算量约为O(ND),固定D时可简称对N线性。底层矩阵运算改变常数而非这一计数结构;是否可接受仍依规模、批处理与硬件实测,不能概括大型语料绝不可能使用。
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
How can you do semantic search on 100D vectors without an index? You now know how to convert the query string into a topic vector using LSA, and you know how to compare two vectors for similarity using the cosine similarity score (the scalar product, inner product, or dot product) to find the closest match. To find precise semantic matches, you need to find all
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:摘录中关于穷举检索复杂度包含维数的定义、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:How can you do semantic search on 100D vectors without an index? You now know how to convert the query string into a topic vector using LSA, and you know how to compare two vectors for similarity using the cosine similarity score (the scalar product, inner product, or dot product) to find the closest match. To find precise semantic matches, you need to find all