一种基于协同过滤的旅游行程推荐算法  

Travel Recommendation Algorithm Based on Collaborative Filtering

作　　者：方潇[1] 刘晓寒 柴永平 周文曼 

机构地区：[1]信息工程大学地理空间信息学院 [2]92292部队 [3]东华理工大学地球科学学院

出　　处：《地理空间信息》2016年第7期53-56,共4页Geospatial Information

基　　金：国家高技术研究发展计划资助项目(2012AA12A404);国家自然科学基金资助项目(41171353)

摘　　要：针对现有的个性化推荐算法在旅游行程推荐领域应用较少,技术不成熟等缺陷,提出基于一种改进协同过滤算法的行程推荐算法。该算法首先依据偏好度计算方法分析用户对于不同类型旅游目的地的偏好程度,形成用户-旅游目的地偏好矩阵,并基于偏好矩阵优化K-Means算法的初始聚类中心,为目标用户建立邻近用户集,最后利用改进的相似度计算方法为目标用户寻找偏好相似的历史用户,从而产生行程推荐。实验结果表明,该算法能够快速搜索较小的邻近用户,得出推荐结果,不仅具有较快的推荐效率,而且具备较好的推荐精准度,在一定程度上为个性化旅游行程推荐服务提供了支撑。Aimed at the defects of existing personalized recommendation algorithms and the few applications in the field of travel recommendation,this paper proposed a travel recommendation algorithm based on improved collaborative filtering.In this algorithm,user preference for different types of tourism destinations was analyzed based on the preference calculation method at first.And then,the user-preference of tourist destinations matrix was formatted.The initial cluster center in K-Means algorithm was optimized by this matrix and the sets of similar users were established.Finally,the paper used the improved similarity calculation method to search the history users of similar preference for target user,and produced the travel recommendation.Experimental results show that the algorithm can get the recommended result with both faster recommended efficiency and better recommendation accuracy,and provide support for personalized travel recommendation service.

关 键 词：个性化推荐 用户偏好 协同过滤算法 

分 类 号：P208[天文地球—地图制图学与地理信息工程]