Transforming user check-in data into graph structure data is a popular and powerful way to analyze users' behaviors in the field of recommendation. Graph-based deep learning methods such as graph embeddings and graph neural networks have shown promising performance on the task of point-of-interest recommendation in recent years. Despite effectiveness, existing methods fail to capture deep graph structural information, leading the suboptimal representations. In addition, they lack the ability of learning the influences of both global preference and user preference on the check-in behavior. To address the aforementioned issues, we propose a general framework based on preference-aware graph diffusion, named PGD. We first construct two types of graphs to represent the global preference and user preference. Then, we apply a graph diffusion process to capture the structural information of the generated graphs, resulting in weighted adjacency matrices. Finally, graph neural network-based backbones are introduced to learn the representations of users and POIs on weighted adjacency matrices. A learnable aggregation module is developed to learn the final representations from global preference and user preference adaptively. Extensive experiments on four real-world datasets demonstrate the superiority of PGD on POI recommendation, compared with the mainstream graph-based deep learning methods.

Image topic representation in social networks is vital for people to get significant and valuable content. However, this task is difficult and challenging due to the complexity of image features. This paper proposes a multifeature complementary attention mechanism for image topic representation named CATR. CATR uses scene-level and instance-level object detection methods to obtain the object information on social networks. Here, the image features are divided into focused features and unfocused features. Focused features are used to learn and express semantic information, while unfocused features are used to filter out noise information in focused feature extraction. The attention mechanism is constructed by combining the object features and the features of the image itself, while the image topic representation in social networks is realized by the complementary attention mechanism. Based on the real image data of Sina Weibo and Mir-Flickr 25K, several groups of comparative experiments are constructed to verify the performance of the proposed CATR by leveraging different evaluation measures. The experimental results demonstrate that the proposed CATR obtains an optimal accuracy and significantly outperforms the other comparison methods in image topic representation.