Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

Scalable Trajectory-User Linking with Dual-Stream Representation Networks

Authors: Hao Zhang, Wei Chen, Xingyu Zhao, Jianpeng Qi, Guiyuan Jiang, Yanwei Yu

AAAI 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on check-in mobility datasets from three real-world cities and the nationwide U.S. demonstrate the superiority of Scale TUL over state-of-the-art baselines for large-scale TUL tasks.
Researcher Affiliation Academia 1Faculty of Information Science and Engineering, Ocean University of China, Qingdao, China 2The Hong Kong University of Science and Technology (Guangzhou) EMAIL,EMAIL, EMAIL
Pseudocode No The paper describes the methodology using textual explanations and a block diagram (Figure 1), but does not contain any clearly labeled pseudocode or algorithm blocks.
Open Source Code Yes The source code of our model is available at https://github.com/sleevefishcode/Scalable TUL.
Open Datasets Yes We use real-world check-in data collected from the popular location-based social network platform Foursquare (Yang et al. 2015), selecting data from three cities and the entire United States for our dataset.
Dataset Splits Yes In our experiments, we use the first 80% of each user s sub-trajectories for training, and the remaining 20% for testing. Additionally, 20% of the training data is set aside as a validation set to assist with an early stopping mechanism to find the best parameters and avoid overfitting.
Hardware Specification Yes All experiments are conducted on a machine with Intel(R) Xeon(R) Silver 4214 (2.20GHz 12 cores) and NVIDIA Ge Force RTX 3090 (24GB Memory).
Software Dependencies No The paper does not explicitly state specific software dependencies with version numbers. It mentions models like RNN, Transformer, Structured State Space Models, and Bi-directional Long Short-Term Memory but not the software libraries or their versions used for implementation.
Experiment Setup Yes For Scale TUL, we set default embedding dimension to 512, apply an early stopping mechanism with patience to 5 to avoid over fitting, and adjust the learning rates as follows: In the first stage, the initial learning rate is set to 0.001 and decays by 20% every 5 epochs. In the second stage, the initial learning rate is set to 0.0005 and decays by 90% every 5 epochs.