基于深度表示模型的旅游路线推荐算法研究  

Travel Route Recommendation Algorithm Based on Depth Representation Model

作　　者：梁栋屹 陶宏曜 刘兴旺[1] 

机构地区：[1]中南民族大学计算机科学学院,湖北武汉430074

出　　处：《软件导刊》2017年第12期158-161,共4页Software Guide

基　　金：大学生创新创业资助项目(GCX16013)

摘　　要：针对现有旅游路线推荐算法在实际应用中景点标签缺失、依赖用户评分等缺陷,提出一种基于深度表示模型的旅游路线推荐算法。根据时空轨迹隐含的位置顺序和时间中的用户移动模式,建立深度表示模型对每个用户的时空间序列训练特征向量,并用Kmeans算法对训练结果进行聚类。为验证实验结果,将用户的移动模式应用到旅游路线推荐上,选择使用Gowalla上的签到数据集进行检测。实验结果表明,包含诸如"购物","夜生活"等标签的移动轨迹具备推荐价值。In currently,there are some weakness like lacking of feature spot label and the large distance between different places for the travel route recommendation algorithm.In this paper,we analyze the sequence of locations and time which hide the user moving model and create a depth representation model to train the feature vector about the space-time list.At last,we select the Kmeans algorithm to cluster the value of feature.To test the result of experiment and apply it in the recommendation of travel route,I select to use the check-in data list Gowalla as the resource.According to the result,some moving route including shopping or night life label have the value of recommendation.

关 键 词：社交网络 深度表示模型 Kmeans聚类 用户移动模式 训练特征向量 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]