摘要
在Web2.0时代,随着智能手机、数码相机和GPS导航系统等电子产品的广泛普及和社交网站的迅速发展,涌现出各种UGC(User Generated Content)形式的数据。同时,人们喜欢以图片或文字方式在网络上分享自己旅游的所见所闻,社交媒体数据通常包括文本标签、地理位置(经纬度)和拍摄时间等信息,这就为研究旅游推荐提供了可靠数据。使用Flickr网站上Geo-tagged照片数据集,采用基于密度的DBSCAN聚类算法对照片的经纬度进行聚类,结合TF-IDF算法为兴趣点命名,得到游客在西安的旅游兴趣点,然后综合考虑用户对兴趣点偏好和兴趣点属性,利用改进的协同过滤推荐算法为用户提供旅游推荐服务。实验结果表明,该算法能够有效提高系统的推荐精度。最后构建了用户信任网络,提高了推荐系统的信任度和满意度。
In the Web2.0 era, with the popularity of smart phones, digital cameras and GPS navigation systems and other portable electronic products widely available and the rapid development of social network,all kinds of UGC （ User Generated Content） are emerging by the social networking sites. Meanwhile, more and more tourists tend to share their travel seen and heard on the network with pictures or texts, and those social media data usually contain textual labels, spatial location （in terms of latitude and longitude）, taken time and other infortnation, which provide truly reliable data. Therefore, the Geo-tagged photo from Flickr is used as data sources, applying the density-based clustering algorithm DBSCAN to cluster latitude and longitude of photos,and getting Points Of Interest （POIs） in Xi＇ an with TF -IDF algorithm. The travel recommendation is provided using improved collaborative filtering algorithm, which considers both user preferences and attributes of POI. The results show that it can improve the recommendation accuracy effectively. Finally the trust network for users is built to improve the trust and satisfaction of the recommendation system.
出处
《计算机技术与发展》
2016年第10期123-126,132,共5页
Computer Technology and Development
基金
国家自然科学基金资助项目(41271387)
