推荐算法：基于图的算法

上图：用户u到物品i的边，表示用户u对物品i进行了评分。给予边权值，表示u对i的评分
@@@a random-walk based scoring algorithm for recommender engines
@@@ applying associative retrieval techniques to alleviate the sparsity problem in collaborative filtering

• 两种推荐方式：

  1. 图中的用户u对物品i的相近距离直接用于推荐。找用户u在图中的最近物品
     @@@random-walk computation of similarities between nodes of a graph with application to collaborative recommendation
     @@@a random-walk based scoring algorithm for recommender engines
     @@@ applying associative retrieval techniques to alleviate the sparsity problem in collaborative filtering

  2. 将用户间或者物品间的相似距离看成他们之间的相似度权重$w_{uv}$或者$w_{ij}$,然后使用邻域的方法
     @@@random-walk computation of similarities between noedes with application to collaborative recommendation
     @@@ a collabotive filtering framework based on both local user similarity and global user similarity

---

基于路径的相似度

最短路径

@@@Horing hatches an egg: A new graph-theoretic approach to collaborative filtering
* 通过用户结点最短距离来计算他们之间相似度的推荐方法

路径数量

@@@ applying associative retrieval techniques to alleviate the sparsity problem in collaborative filtering
计算路径数量来估测连通度

---

随机游走模型

概率框架

itemrank

@@@a random-walk based scoring algorithm for recommender engines

---

平均首次通过/往返次数

• 针对随机游走的测量距离
  @@@random-walk computation of similarities between noedes with application to collaborative recommendation
  @@@An Emperimental investigation of graph kernels on a collaborative recommendation task.