Skip to content

Commit

Permalink
fix: 4.6.6.2.3 latex
Browse files Browse the repository at this point in the history
  • Loading branch information
camera-2018 committed Apr 20, 2023
1 parent 885a01d commit 655f069
Showing 1 changed file with 5 additions and 16 deletions.
21 changes: 5 additions & 16 deletions 4.人工智能/4.6.6.2.3序列化推荐.md
Original file line number Diff line number Diff line change
Expand Up @@ -2,24 +2,13 @@

# 什么是序列化推荐?

在现实世界中,用户的前后行为都存在强烈的关联性与因果性,将用户过去的这一系列交互行为视作用户行为序列 u={
$$
i_1
$$
在现实世界中,用户的前后行为都存在强烈的关联性与因果性,将用户过去的这一系列交互行为视作用户行为序列 u={$i_1$,$i_2$,……,$i_n$}并通过构建模型对其建模,来预测下一时刻用户最感兴趣的内容$i_{n+1}$,这就是序列化推荐(Sequential Recommendation)的核心思想。目前序列化推荐的常用模型包括RNN、CNN、GNN以及Transformer。

,
$$
i_2
$$

,……,$$i_n

$$
}并通过构建模型对其建模,来预测下一时刻用户最感兴趣的内容$$i_{n+1}$$,这就是序列化推荐(Sequential Recommendation)的核心思想。目前序列化推荐的常用模型包括RNN、CNN、GNN以及Transformer。
> 传统的推荐系统,例如基于内容和协同过滤的推荐系统,以一种静态的方式建模用户和商品的交互并且只可以捕获用户广义的喜好。<br/><br/>而SRSs则是将用户和商品的交互建模为一个动态的序列并且利用序列的依赖性来活捉当前和最近用户的喜好。
> 传统的推荐系统,例如基于内容和协同过滤的推荐系统,以一种静态的方式建模用户和商品的交互并且只可以捕获用户广义的喜好。
<br/><br/>
而SRSs则是将用户和商品的交互建模为一个动态的序列并且利用序列的依赖性来活捉当前和最近用户的喜好。

![](static/boxcnolggxKhDZDBzIFPIaDFfhc.png)
$$

0 comments on commit 655f069

Please sign in to comment.