基于多任务权重参数学习的兴趣点推荐算法研究

Research on Interest Point Recommendation Algorithm Based on Dynamic Weight Parameter Learning

基于位置的社交网络的快速发展,通过借助社交网络来分享用户位置信息,其中包含了丰富的上下文信息,比如用户签到、兴趣点地理位置、用户社交关系等,从而对兴趣点推荐的发展提供了很大的机遇。但是,如何有效地利用这些上下文信息,解决数据稀疏和隐式反馈等问题,是很大的挑战。针对这一问题,提出了一种能够动态融合不同上下文因素的推荐算法,该推荐算法可以融合不同类型的上下文因素,比如地理信息、类别信息、时间信息等,通过一种类似梯度下降的动态权重参数学习的方法,动态地学习每个因素的权重,适应不同类型用户特点,从而改善兴趣点推荐效果。

With the rapid development of location-based social networks,users'location information is shared by means of social networks,which contains rich contextual information,such as user check-in,geographical location of interest points,and user social relationship,which provides a great opportunity for the development of interest point recommendation.However,how to effectively use the context information to solve the problems of data sparsity and implicit feedback is a great challenge.To solve this problem,This paper proposes a recommendation algorithm that can dynamically fuse different context factors.The recommendation algorithm can fuse different types of context factors,such as geographic information,category information,time information,etc.through a dynamic weight parameter learning method similar to gradient descent,the weight of each factor can be dynamically learned to adapt to the characteristics of different types of users,so as to improve Interesting point recommendation effect.