Point of interests recommendation in location-based social networks

Abstract

With the rapid development of location-based social networks (LBSNs), point of interests (POI) recommendation has become an important means to help people discover attractive and interesting locations from billions of locations globally. However, this recommendation is very challenging compared to the traditional recommender systems. A user may visit only a limited number of POIs, leading to a very sparse user-item matrix. This matrix becomes even sparser when the user travels to a distant place as most of the items visited by a user are usually located within a short distance from the user's home. Moreover, user interests and behavior patterns may vary dramatically across different time and different geographical regions. On the other hand, in reality, human movement exhibits sequential patterns. Thus, how to predict users' next move based on her previous visited locations is important and challenging in LBSNs. Our project focuses on offering a more accurate and efficient recommender system by overcoming the aforementioned challenges, and it contains the following three parts:In the first part, we design ST-SAGE, a spatial-temporal sparse additive generative model for POI recommendation. ST-SAGE considers both personal interests of the users and the preferences of the crowd in the target region at the given time by exploiting both the co-occurrence patterns of POIs and the content of POIs. To further alleviate the data sparsity issue, ST-SAGE exploits the geographical correlation by smoothing the crowd's preferences over a well-designed spatial index structure called spatial pyramid. To speed up the training process of ST-SAGE, we implement a parallel version of the model inference algorithm on the GraphLab framework.The second part aims to leveraging sequential patterns in POI recommendation. However, this is very challenging, considering 1) users' check-in data in LBSNs has a low sampling rate in both space and time, which renders existing location prediction techniques on GPS trajectories ineffective;2) the prediction space is extremely large, with millions of distinct locations as the next prediction target, which impedes the application of classical Markov chain models; and 3) there is no existing framework that unifies users' personal interests and the sequential influence of recently visited locations in a principled manner. In light of the above challenges, we propose a sequential personalized POI recommendation framework (SPORE) which introduces a novel latent variable topic-region to model and fuse sequential influence with personal interests in the latent and exponential space. The advantages of modeling the sequential effect at the topic-region level include a significantly reduced prediction space, an effective alleviation of data sparsity and a direct expression of the semantic meaning of users' spatial activities. Helen Huang. I also want to thank the PhD students I worked with. I really appreciate having had the opportunity to work with them, and their friendships will always be reme...

Behavior‐based POI recommendation for small groups in location‐based social networks

“…First, a power-law (PL) distribution is the probability of a user moving from one place to another with the geographical distance function (Ma et al, 2020;Wang, 2017;Ye, Yin, Lee, & Lee, 2011). Second, the multi-center Gaussian model (MGM) is based on users' tendency to check-in around multiple centers (Cheng et al, 2012(Cheng et al, , 2016.…”

“…Finding POIs that satisfy most group members with various distances is a challenge. There are many studies about individual recommender systems, which have considered geographical influence (Cheng, Yang, King, & Lyu, 2012Christoforidis et al, 2019;Christoforidis, Kefalas, Papadopoulos, & Manolopoulos, 2021;Kefalas & Manolopoulos, 2017;Kefalas, Symeonidis, & Manolopoulos, 2016, 2018Khoshahval, Farnaghi, Taleai, & Mansourian, 2018;Liu, Zhang, Liu, Qiu, & Zhang, 2020;Ma et al, 2020;Rahimi, Far, & Wang, 2020;Wang, 2017;Xie et al, 2016). Due to the group complexity, they cannot be used in making suggestions to a group.…”

“…Like other preferences in this strategy, making an individual recommendation using methods that consider the geographical influence and aggregating the recommendations is one solution for applying the geographical influence in a group recommendation (Cheng et al, 2012(Cheng et al, , 2016Liu et al, 2020;Ma et al, 2020;Wang, 2017).…”

A Personalized Hybrid Recommendation Algorithm for Location-Based Service on Smart Campus

“…In recent years, researchers have conducted a number of recommendation algorithms based on user's geographic locations. For example, in [4], the cross adjustment is conducted to achieve the convergence of user similarity and location similarity establishing a user-tendency model and calculating the similarity, and the personalized recommended list is generated based on both users' interest and the distance of recommendation locations; Yuichiro Takeuchi and Masanori Sugimoto explore stores that are frequently visited by visitors through the analysis of the user's GPS history information, and take these stores as the input of the item-based collaborative filtering algorithm [5]. With the development of positioning technology and the popularization of social networks, the moving trajectories of users' geographical location are closely related to the laws of social life.…”