Next POI Recommendation with Dynamic Graph and Explicit Dependency

Authors: Feiyu Yin, Yong Liu, Zhiqi Shen, Lisi Chen, Shuo Shang, Peng Han

AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate the proposed model on two real-world datasets, and the experimental results show that the proposed method significantly outperforms existing state-of-the-art POI recommendation methods.
Researcher Affiliation Academia 1 University of Electronic Science and Technology of China 2 Nanyang Technological University, Singapore 3 Sichuan Artificial Intelligence Research Institute, Yibin, 644000, China
Pseudocode No The paper does not contain structured pseudocode or an algorithm block.
Open Source Code Yes Our implementation is available in Pytorch3. 3https://github.com/Shirley-YFY/SNPM
Open Datasets Yes The experiments are conducted on two widely used real world datasets: Gowalla1 and Foursquare2. 1http://snap.stanford.edu/data/loc-gowalla.html 2https://sites.google.com/site/yangdingqi/home
Dataset Splits No The first 80% check-ins of each user are split into multiple length-equally (e.g., 20) sequences, which are used as training set, Likewise, the remaining 20% check-ins are used as testing set. The paper does not explicitly mention a separate validation dataset split.
Hardware Specification Yes The experiments are performed in the environment with the following hardware platform: CPU: AMD Ryzen 5 3600, GPU: NVIDIA Ge Force RTX 3090TI.
Software Dependencies No The paper mentions 'Pytorch' for implementation but does not specify its version or any other software dependencies with version numbers.
Experiment Setup Yes When constructing the POI transition graph through Rotat E, we choose 100 neighbors for each POI. When using spectral clustering for POI clustering, we choose the 20 smallest eigenvalues and the corresponding eigenvectors, and set ω = 10 in SDNG. In the prediction Model, ρ is set to 0.05, κ is set to 1/32. Moreover, the L2 regularization coefficient β is set to 1.5e 6 and 2e 7 on Gowalla and Foursquare datasets, respectively. In MSRNN, the number of historical hidden state representations that are explicitly used is 6