tf-idf weighting

水平有限，如有错误请指正！

tf-idf weighting

tf(term frequency)

a document or zone that mentions a query term more often has more to do with that query and therefore should receive a higher score
query term: a set of words
$N$:document中term的总数
$t$:document中term出现的次数

$$tf_t=\frac{t}{N}$$

idf(inverse document frequency)

$N$: document的数量
$df_t$:有多少documents 包含$t$

$$idf_t=log\frac{N}{df_t}$$
从公式可以看出：

• $df_t$ 越小，$idf_t$越大,表明$t$对文档的区分性更大
• 反之

tf-idf weighting

$$\text{tf-idf}_{t,d}=tf_{t,d}* idf_t$$

参考：
tf
idf
tf-idf