知识卡片
局部词面匹配可能选错问题意图
decrease等共享词可能使一个关于boosting的问题超过真正关于Logistic Regression的问题。词项相似并不理解操作对象和条件;应观察候选及错例,再考虑字符特征、领域处理或语义表示,不能靠增加任意数据量保证修好。
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
If you looked closely at the dataset, you might be able to find an answer about ways to decrease overfitting; however, this vectorizer is just a bit too literal. When it saw the word decrease in the wrong question, that caused it to compute a larger dot product with higher similarity than for the correct question. To explore the dataset a bit more thoroughly, you
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第03章《词频与逆文档频率向量》(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-3.html)
- 结论依据:摘录中关于局部词面匹配可能选错问题意图的定义、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:If you looked closely at the dataset, you might be able to find an answer about ways to decrease overfitting; however, this vectorizer is just a bit too literal. When it saw the word decrease in the wrong question, that caused it to compute a larger dot product with higher similarity than for the correct question. To explore the dataset a bit more thoroughly, you