基于知识图谱的旅游推荐系统

论文总字数：18504字

摘 要

随着互联网的高速发展以及人们生活水平的提高，旅游出行已经成为人们享受生活的主要途径之一，并且这种消费需求将进一步扩大^[1]。互联网虽然为游客查询景点信息提供了便利，但依然浪费了游客大量的时间精力。随着大数据时代的到来，繁杂的数据开始爆发，推荐系统的准确度低，以及数据存储等问题越来越严重，如何提高用户的满意度以及推荐系统的准确度，是当前推荐系统面临且必须解决的问题^[2]。知识图谱的出现为解决这一问题提供了极大便利，它作为一种能够融合多源异构数据，从“关系”角度分析问题，对数据进行深度挖掘，将自然语言转换成计算机语言，最大限度的展示数据价值，可以应用于智能推荐、智能问答、智能搜索等领域^[3]。本课题就是探讨如何基于知识图谱的深度学习模型的推荐算法给用户推荐路线的。研究内容如下：

1.本文手动从携程网、去哪儿网等网站爬取南京旅游景点宾馆用户数据,并对多源数据进行信息抽取，知识融合加工，整合处理后导入Neo4j图数据库^[4],构建出了南京旅游知识图谱。

2.在南京旅游知识图谱的基础上，构建和训练基于深度学习模型的推荐算法：本文采用深度学习模型Node2Vec实现Graph Embedding（图嵌入）。通过随机游走的方法对构建好的南京旅游知识图谱进行实体序列采样，生成序列样本；根据神经网络模型对序列样本进行建模，得到图谱各个实体的特征，通过计算各个实体之间的余弦相似度，得到最后的推荐列表。

本文在旅游景点知识图谱的基础上，提出了将宾馆这一实体加入到知识图谱的构建中，同时加入了游客与宾馆的入住、景点与宾馆的相距语义关系。通过深度学习方法将基于游客、景点和宾馆三个主要实体构建的知识图谱，映射到同一个向量空间中，可以同时计算景点、宾馆和游客之间的向量相似度，从而进一步丰富了旅游知识图谱的语义信息。

关键词：知识图谱；推荐系统；Neo4j图数据库；深度学习；word2vec模型；图嵌入

Travel recommendation system based on knowledge graph

Abstract

With the rapid development of the Internet and the improvement of people's living standards, travel has become one of the main ways for people to enjoy life, and this consumer demand will be further expanded [1]. Although the Internet provides convenience for tourists to query information about scenic spots, it still wastes a lot of time and energy for tourists. With the advent of the era of big data, complicated data has begun to explode, the accuracy of the recommendation system is low, and the problems of data storage are becoming more and more serious. How to improve user satisfaction and accuracy of the recommendation system is currently facing and Problems that must be solved [2]. The emergence of knowledge graph provides great convenience for solving this problem. As a kind of fusion of multi-source heterogeneous data, it analyzes the problem from the perspective of “relationship”, conducts deep mining on the data, and converts natural language into computer language The limited display of data value can be applied to the fields of intelligent recommendation, intelligent question answering, and intelligent search [3]. This topic is to discuss how to recommend routes to users based on the recommendation algorithm of deep learning model of knowledge graph. The research content is as follows:

1. This article manually crawls the user data of Nanjing tourist attractions hotels from websites such as Ctrip.com, Qunar.com, and extracts information from multi-source data, processes knowledge fusion, and imports it into the Neo4j graph database [4] after construction. Atlas of tourism knowledge.

2. On the basis of Nanjing Tourism Knowledge Graph, construct and train a recommendation algorithm based on deep learning model: This article uses deep learning model Node2Vec to implement Graph Embedding (graph embedding). The random travel method is used to sample the constructed Nanjing tourism knowledge map for entity sequence sampling to generate sequence samples; the sequence samples are modeled according to the neural network model to obtain the characteristics of each entity in the map, and the cosine similarity between each entity is calculated Degree to get the final recommendation list.

Based on the knowledge graph of tourist attractions, this paper proposes to add the entity of hotel to the construction of the knowledge graph, and at the same time it adds the semantic relationship between tourists and hotels, and the distance between attractions and hotels. The knowledge map based on the three main entities of tourists, attractions and hotels is mapped into the same vector space through deep learning methods, and the vector similarity between attractions, hotels and tourists can be calculated at the same time, thereby further enriching the tourism knowledge map Semantic information.

Key Words: Knowledge graph; recommendation system; Neo4j graph database; deep learning; word2vec model; graph embedding

目录

摘要 I

Abstract II

第一章 绪引 5

1.1 系统研究背景及意义 5

1.1.1 背景 5

1.1.2 意义 5

1.2 研究现状 5

1.3 系统研究内容 6

1.4 论文组织结构 6

第二章 相关技术基础 7

2.1 知识图谱简介 7

2.1.1知识图谱概述 7

2.1.2知识图谱构建框架设计 7

2.2 荐系统介绍 8

2.3 数据库 8

2.4 深度学习模型的推荐算法 8

2.5 本章小结 8

第三章 推荐系统需求分析 9

3.1 系统功能需求分析 9

3.2 系统非功能需求分析 9

第四章 系统设计 10

4.1系统总体设计 10

4.1.1 系统分层结构设计 10

4.1.2 系统功能模块结构设计 10

4.1.3 系统总体工作流程设计 11

4.2 统详细设计 11

4.2.1 知识图谱构建 11

4.2.2 图数据库设计 15

4.2.3 旅游知识图谱显示 16

4.3基于知识图谱的景点和宾馆推荐算法 18

4.3.1基于Node2Vec模型的算法原理 18

4.3.2基于Node2Vec模型构建推荐系统的流程 19

4.3.3 随机游走（Random Walk）对图谱进行实体序列采样 19

4.3.4 Word2Vec方法将序列样本映射成向量 20

第五章 系统实现与测试 21

5.1系统实现工具与环境 21

5 .2 核心代码分析 21

剩余内容已隐藏，请支付后下载全文，论文总字数：18504字