文献类型：期刊文献

英文题名：CLGENRec: Contrastive Learning and Graph Enhanced Network for Next POI Recommendation

作者：Zhu, Yi[1]; Guo, Weibin[1]

机构：[1] East China University of Science and Technology, School of Information Science and Engineering, Shanghai, 200237, China

年份：2024

起止页码：265

外文期刊名：2024 IEEE 4th International Conference on Electronic Technology, Communication and Information, ICETCI 2024

收录：EI(收录号：20243216802914)

语种：英文

外文关键词：Learning systems - Long short-term memory - Marketing - Telecommunication services - User profile

摘要：With the rapid development of location-based services, next point-of-interest (POI) recommendation has become a prominent research direction in the field of geographic location services. Effectively predicting users' next visit interests can enhance the user experience and provide accurate marketing strategies for businesses. However, data sparsity and the cold start are significant challenges in next POI recommendation. To address the sparsity issue, we first obtain the user's stable preference and dynamic preference by common learning trajectory features through long-term and short-term modeling. We use an LSTM that considers periodicity and weight attenuation to learn dynamic features to enhance user embedding, so as to obtain richer contextual features to alleviate data sparsity. Secondly, we can better distinguish the user's preferences and non-preferences by introducing the method of contrastive learning, so as to enhance the data. To tackle the cold start problem, the global trajectory graph is introduced for guiding the learning of the model. Consequently, we proposes a Contrastive Learning and Graph Enhanced Network for Next POI Recommendation (CLGENRec). Finally, the experiments on three datasets proves the superiority of this model. ? 2024 IEEE.