融合时间序列的POI动态推荐算法

兴趣点(POI)的签到数据体现了用户的偏好和兴趣点的分布特征,这在兴趣点推荐领域有极为重要的价值.为了缓解数据稀疏造成的推荐不准确等问题,本文提出了融合时间序列的POI动态推荐算法,结合用户与用户之间的关系、兴趣点位置以及流行度信息等.首先划分时间序列,得到时间因子的相似度;其次时间序列融入到基于用户的协同过滤算法,再根据时间的连续性特征得到基于用户的预测评分,然后将地理影响因子与基于时间的流行度信息结合,预测用户的评分,进而与基于用户的评分加权融合;最后,在Gowalla数据集上进行实验,结果表明,本文提出的融合时间序列的POI动态推荐算法能够有效减小推荐误差,提高推荐精度与召回率.

POI Dynamic Recommendation Algorithm Based on Fusion Time Series

The check-in data of point of interest(POI)reflects the distribution characteristics of the user’s preferences and points of interest,which is of great value in the field of point of interest recommendation.In order to alleviate the problem of inaccurate recommendation caused by data sparseness,this paper proposes a POI dynamic recommendation algorithm that combines time series,combining the relationship between users and users,the location of interest points and popularity information.Firstly,the time series is divided to obtain the similarity of time factors.Secondly,the time series is integrated into collaborative filtering algorithm based on users.Then,based on the continuity characteristics of time,the prediction scores based on users is obtained,and then based on the geographical influence factors popularity information to predict user ratings and then weight-based fusion with user-based ratings.In the end,the experiments are carried out on the Gowalla dataset.The experimental results show that the proposed POI dynamic recommendation algorithm for fusion time series can effectively reduce the recommendation error and improve the recommendation accuracy and recall.