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. Our code is available at <a class="link-external link-https" href="https://github.com/Yukayo/LoTNext" rel="external noopener nofollow">this https URL</a>.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the **long - tail problem in human mobility prediction**. Specifically, the paper focuses on how to accurately predict the location (i.e., POI, point of interest) that a user will visit next, especially those long - tail POIs with low visit frequencies. This type of problem is very important in practical applications because: 1. **Personalized navigation and recommendation system optimization**: By predicting the user's next POI, more personalized services can be provided to the user, such as navigation, recommending restaurants or scenic spots, etc. 2. **Urban traffic planning and environmental protection**: Accurately predicting the flow of people can help urban planners better manage traffic flow, reducing congestion and pollution. However, existing AI models face challenges when dealing with long - tail POIs, mainly due to the imbalance of data distribution. Most users tend to visit popular locations and visit less - popular locations less frequently, resulting in an obvious long - tail distribution of data. This imbalance makes it difficult for the model to learn the characteristics of long - tail POIs, thus affecting the prediction performance. To address this problem, the paper proposes the **Long - Tail Adjusted Next POI Prediction (LoTNext)** framework, which aims to improve the prediction effect of long - tail POIs through the following methods: 1. **Long - Tailed Graph Adjustment Module**: - By adjusting the long - tail nodes in the user - POI interaction graph, the impact of these nodes on the model performance is reduced. - Specifically, this module will evaluate the importance of edges through the attention mechanism and remove those low - quality edges, thereby retaining high - quality interaction information. 2. **Long - Tailed Loss Adjustment Module**: - By adjusting the logit scores and sample weights in the loss function, the loss between head - class and tail - class samples is balanced. - For example, for frequently occurring POIs (head - class), the impact on the loss is reduced; while for less - occurring POIs (tail - class), the impact on the loss is increased. 3. **Auxiliary Prediction Task**: - An additional task (such as time prediction) is introduced to enhance the generalization ability and accuracy of the model without relying on additional data sources. Through these methods, LoTNext can more effectively utilize the information of long - tail POIs, improving the accuracy and robustness of prediction. The experimental results show that LoTNext significantly outperforms the existing state - of - the - art methods on two real - world data sets, especially in predicting long - tail POIs.

Taming the Long Tail in Human Mobility Prediction

Space-Time Adaptive Network for Origin-Destination Passenger Demand Prediction

Lifting the Predictability of Human Mobility on Activity Trajectories

Physics-Informed Spatio-Temporal Model for Human Mobility Prediction

nextlocllm: next location prediction using LLMs

Human mobility prediction with causal and spatial-constrained multi-task network

Human Mobility Prediction Using Sparse Trajectory Data.

Exploring the potential of deep regression model for next-location prediction

Towards Effective Next POI Prediction: Spatial and Semantic Augmentation with Remote Sensing Data

Where Would I Go Next? Large Language Models as Human Mobility Predictors

Individual Mobility Prediction via Attentive Marked Temporal Point Processes

Predicting Human Mobility Via Variational Attention.

Location Prediction via Bi-direction Speculation and Dual-level Association

Mobility Prediction Via Sequential Trajectory Disentanglement (student Abstract).

Activity-aware Human Mobility Prediction with Hierarchical Graph Attention Recurrent Network

A Systematic Analysis of Fine-Grained Human Mobility Prediction with On-Device Contextual Data

Human Mobility Prediction from Region Functions with Taxi Trajectories

Prediction of Urban Human Mobility Using Large-Scale Taxi Traces and Its Applications

Context-aware multi-head self-attentional neural network model for next location prediction

MobTCast: Leveraging Auxiliary Trajectory Forecasting for Human Mobility Prediction

Attentional Markov Model for Human Mobility Prediction