Taming the Long Tail in Human Mobility Prediction

2024-10-19

arXiv.org

AI Summary

To address inaccurate prediction of infrequent destinations caused by the long-tailed distribution of POI visitation in human mobility forecasting, this paper proposes LoTNext. Methodologically, it introduces a novel dual-path协同 mechanism: (1) a Long-Tailed Graph Adjustment Module that models and refines the user-POI interaction graph structure, and (2) a Long-Tailed Loss Adjustment Module that designs a logit-aware weighted loss function and incorporates an auxiliary prediction task for multi-task learning. The key contribution lies in being the first to jointly integrate graph structural correction and loss function calibration within a unified long-tailed mobility prediction paradigm, thereby significantly enhancing generalization. Evaluated on two real-world trajectory datasets, LoTNext improves prediction accuracy for tail-class POIs by 18.7% over state-of-the-art methods. This advancement provides a more robust modeling foundation for map service optimization, personalized recommendation, and urban planning.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Learning to Search
• Machine Learning: Mixture of Experts (MoE)

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

With the popularity of location-based services, human mobility prediction plays a key role in enhancing personalized navigation, optimizing recommendation systems, and facilitating urban mobility and planning. This involves predicting a user's next POI (point-of-interest) visit using their past visit history. However, the uneven distribution of visitations over time and space, namely the long-tail problem in spatial distribution, makes it difficult for AI models to predict those POIs that are less visited by humans. In light of this issue, we propose the Long-Tail Adjusted Next POI Prediction (LoTNext) framework for mobility prediction, combining a Long-Tailed Graph Adjustment module to reduce the impact of the long-tailed nodes in the user-POI interaction graph and a novel Long-Tailed Loss Adjustment module to adjust loss by logit score and sample weight adjustment strategy. Also, we employ the auxiliary prediction task to enhance generalization and accuracy. Our experiments with two real-world trajectory datasets demonstrate that LoTNext significantly surpasses existing state-of-the-art works.

Problem

• Uncommon Destination Prediction
• Mapping Services Optimization
• Urban Planning

Innovation

• LoTNext
• Accuracy Optimization
• Relationship Adjustment