随着智能手机与照片分享网站如 Flickr 和 Panoramio 等流行，人们可以随时随地拍摄并分享自己的照片。这些带有地理标签的照片能够反映用户的旅游历史，可以用来作为 个性化旅游景点推荐的数据源。这篇论文提出了一种矩阵分解结合用户相似性的旅游景点 推荐方法，首先利用矩阵分解对原始的用户-旅游景点矩阵进行补全，然后基于补全的矩 阵建立用户-用户相似性矩阵以实现个性化旅游景点推荐。本论文在包含了在南京拍摄的 地理标注照片的 Flickr 数据集上进行实验，实验结果证明了矩阵分解结合用户相似性的 方法在旅游景点推荐的准确性方面的有效性。本论文还实现了一个推荐南京市内的旅游景 点的推荐系统原型。 74260

毕业论文关键词 地理标注照片 个性化 旅游景点推荐 矩阵分解 协同过滤 

毕 业 设 计 说 明 书 外 文 摘 要

Title Research on Personalized Recommendation Method  Based on Mining Geotagged Data               

Abstract With the wide use and rapid growth of smart phones and photo sharing websites such as Flickr and Panoramio, people can take photos and share the photos that they have taken anytime and anywhere。 These publicly available geotagged photos can reflect users’ travel history and can be used as the data source of personalized travel location recommendation。 In this paper, we propose a method which combines matrix factorization with user similarity to recommend travel locations。 In the first place, we decompose the original user - travel location preference matrix and end up with a completed user - travel location rating matrix。 After that, we establish the user - user similarity matrix based on the completed user - travel location rating matrix。 At last, we find topK users that are most similar to the target user and then use collaborative filtering which can score and rank travel locations to recommend travel locations which are personalized to the target user。 This paper conducts several experiments on a Flickr data set which contains a large number of geotagged photos which were taken in Nanjing。 The experimental results show that the proposed method which combines matrix factorization with user similarity can effectively recommend travel locations that are personalized to the target user with respect to the precision of travel location recommendation。 We also implement a prototype of recommender system to recommend travel locations in Nanjing。 

Keywords  Geotagged photos   Personalized   Travel location recommendation Matrix factorization     Collaborative filtering 

本科毕业设计说明书 第 I 页

目 次

1  引言（或绪论） 。 1 

1。1  国内外研究情况 。 2 

1。2  概念与问题定义 。 4 

2  方法 。 6 

2。1  发现旅游景点 。 6 

2。2  描述旅游景点名称 。 7 

2。3  描述旅游景点流行参观情境 。 9 

2。4  建立用户偏好矩阵  11 

2。5  矩阵分解  12 

2。6  建立用户-用户相似性矩阵 。 13 

2。7  旅游景点推荐  14 

2。8  推荐系统原型  15 

3  实验  18 

3。1  数据集  18 

3。2  发现旅游景点  18 

3。3  描述旅游景点名称  19 

3。4  描述旅游景点流行参观情境  19 

3。5  矩阵分解  20