08_Graph_and_Network.md

8. Graph and Network

Attentions: This is about Graph, Network, NOT Image, Picture, Computer Vision, etc.

8.1 Overview

Paper

• Learning Representations of Graph Data -- A Survey - UCL2019

  综述：总结和探讨图数据表征学习方法的最新进展。

• https://github.com/thunlp/GNNPapers

  Must-read papers on GNN

• https://github.com/deepgraphlearning/literaturedl4graph

  Paper list about deep learning for graphs 涉及：Node Representation, KG Embedding, GNN, Application, Graph Generation.

• https://github.com/benedekrozemberczki/awesome-graph-classification

  A collection of important graph embedding, classification and representation learning papers with implementations.

• https://github.com/talorwu/Graph-Neural-Network-Review

  GNN综述

Library

• 「紫禁之巅」四大图神经网络架构 - 2020

Article

• 深度学习中不得不学的 Graph Embedding 方法

• 阿里凑单算法首次公开！基于Graph Embedding的打包购商品挖掘系统解析 - 2018

8.2 Node2Vec

Node2Vec belongs to Graph Embedding. Detailed info about Graph Embedding is at <>.

node2vec: Scalable Feature Learning for Networks - Stanford2016

node2vec is an algorithmic framework for learning continuous feature representations for nodes in networks. In node2vec, we learn a mapping of nodes to a low-dimensional space of features that maximizes the likelihood of preserving network neighborhoods of nodes.

node2vec主要用于处理网络结构中的多分类和链路预测任务，具体来说是对网络中的节点和边的特征向量表示方法。简单点来说就是将原有社交网络中的图结构，表达成特征向量矩阵，每一个node(人、物或内容等)表示成一个特征向量，用向量与向量之间的矩阵运算来得到相互的关系。

Code

• http://snap.stanford.edu/node2vec/

• https://github.com/aditya-grover/node2vec

Article

• 关于Node2vec算法中Graph Embedding同质性和结构性的进一步探讨

• node2vec: Embeddings for Graph Data

8.3 Graph Neural Network

Paper

• Graph convolutional networks: a comprehensive review - 2019

• Graph Convolutional Networks for Text Classification

  Article: 论文总结】TextGCN - 2020

Code

• https://github.com/tkipf/gcn (Tensorflow)

Article

• 从图(Graph)到图卷积(Graph Convolution): 漫谈图神经网络 一, 二, 三

• 目前看的GNN论文的一些总结 - 2020

• PinSAGE：GCN 在工业级推荐系统中的应用 - 2020