图计算与推荐系统

　　这是一本全面讲解图计算、知识图谱及其在推荐系统领域应用的专著，为读者基于神经网络构建推荐系统提供了详细指导，是作者在相关领域10余年经验的总结。掌握本书内容，读者可开发出能处理多模态数据的推荐算法系统，提供更丰富和准确的推荐体验。本书主要内容分为两篇。第一篇 图数据以及图模型（第1-3章）对图数据、图神经网络、知识图谱的基础知识进行了梳理，帮助读者掌握着3项技术的关键原理与算法，为后面的学习打下基础。第二篇 推荐系统（第4-9章）首先介绍了推荐系统的架构，包括逻辑架构、技术架构和数据建模，以及基于GNN的推荐系统架构；然后详细讲解了如何基于GNN构建推荐系统，以及基于图的推荐算法；再接着讲解了知识图谱在推荐系统中的应用以及相关的算法模型；最后，探讨了推荐系统领域当前的热点问题、研究方向以及工业级推荐系统领域的核心难题本书注重实战，故理论知识简练且极具针对性，包含大量实战案例，图文并茂，易于阅读。

　　刘 宇资深AI技术专家和搜索与推荐领域专家，曾在多家互联网公司担任资深算法专家、技术总监以及技术VP，现担任某创业公司CTO。在人工智能和信息检索领域有10余年开发经验，对主流的推荐、搜索、聊天机器人、大模型等技术、产品与解决方案都有深入研究，尤其擅长用简单高效的方法解决公司的数智化问题。项目经验丰富，曾成功主导多个电商算法项目的落地和实施，参与完成多个推荐系统从0到1的搭建。曾在多家单位获得个人开发优秀贡献奖，带领团队多次获得团队优秀贡献奖。著有《智能搜索和推荐系统：原理、算法与应用》《聊天机器人：入门、进阶与实战》，其中前者在2022年被某电商平台评为“人工智能领域最受读者喜爱图书ToP5”。

Contents..目　　录
推荐序一
推荐序二
推荐序三
前言
第一篇 图数据与图模型
第1章 图数据基础 ··························2
1.1 数学基础 ·····································2
1.2 图的基本知识 ······························4
1.2.1.什么是图 ·························4
1.2.2.图中基本元素及定义 ·········5
1.3 图的表示方法 ····························10
1.3.1.图的代数表示 ················11
1.3.2.图的遍历 ·······················13
1.4 图数据及图神经网络 ··················14　
1.4.1.图数据的性质 ················14
1.4.2.图数据应用 ···················15
1.4.3.图神经网络的发展史 ·······16
1.5 本章小结 ···································17
第2章 图神经网络基础 ·················18
2.1 神经网络的基本知识 ··················18
2.1.1.神经元 ··························19
2.1.2.前馈神经网络 ················22
2.1.3.反向传播 ·······················23
2.2 卷积神经网络 ····························24
2.2.1.卷积神经网络基本概念
和特点 ··························25
2.2.2.卷积神经网络模型 ··········29
2.3 循环神经网络 ····························30
2.3.1.循环神经网络结构和
特点 ·····························31
2.3.2.循环神经网络模型 ··········35
2.4 图神经网络 ································36
2.4.1.图神经网络综述 ·············36
2.4.2.卷积图神经网络 ·············41
2.4.3.循环图神经网络 ·············42
2.5 本章小结 ···································44
第3章 知识图谱基础 ·····················46
3.1 知识图谱的定义和模型 ···············46
3.1.1.知识图谱定义 ················47
3.1.2.知识图谱嵌入 ················48
3.1.3.距离变换模型 ················51
3.1.4.语义匹配模型 ················53
3.2 知识图谱上的神经网络 ···············55
3.2.1.关系图卷积网络 ·············55
3.2.2.知识图谱与注意力模型 ·····55
3.3 本章小结 ···································59
第二篇 推荐系统
第4章 推荐系统架构 ·····················62
4.1 推荐系统的逻辑架构 ··················62
4.2 推荐系统的技术架构 ··················67
4.3 推荐系统的数据和模型部分 ········69
4.3.1.推荐系统中的数据平台
建设 ·····························69
4.3.2.推荐系统中的数据挖掘
方法 ·····························73
4.3.3.推荐系统模型 ················76
4.4 推荐系统的评估 ·························81
4.4.1.推荐系统的评估实验
方法 ·····························89
4.4.2.离线评估 ·······················89
4.4.3.在线评估 ·······················92
4.5 基于GNN的推荐系统架构 ·········94
4.6 本章小结 ···································96
第5章　基于GNN的推荐系统构
建基础 ·······························97
5.1 关于嵌入 ···································97
5.2 Word2Vec ·································102
5.2.1.哈夫曼树与哈夫曼编