SR-GNN, 图网络召回

简述

将用户的session序列构建成有向图,通过考虑item之间的转移关系, 更好的学习item emb及用户的向量表达.

related work

一些流行的GNN演变.
word2vec, DeepWalk, LINE, node2vec.

结构及解读

网络结构

nodes connection

$A_s$As​ is defined as the concatenation of two adjacency matrices. directed edges have normalized weight.

gated recurrent units

session emb

$hybrid\_session\_emb=linear([s_{local};s_{global}])$hybrid_session_emb=linear([slocal​;sglobal​]), where $s_{global}=soft\_attention(\cdot)$sglobal​=soft_attention(⋅)

experiment

• baselines. GRU4REC, STAMP, etc.
• dataset. YooChoose from RecSys Chanllenge 2015.
• P@20=71.36, evaluation metric , identical to recall@20.

参考

1. paper,AAAI,2019, Session-based Recommendation with Graph Neural Networks
2. corresponding official code,SR-GNN