Abstract:Multi-modal transportation recommendation aims to provide the most appropriate travel route with various transportation modes according to certain criteria. After analyzing large-scale navigation data, we find that route representations exhibit two patterns: spatio-temporal autocorrelations within transportation networks and the semantic coherence of route sequences. However, there are few studies that consider both patterns when developing multi-modal transportation systems. To this end, in this paper, we study multi-modal transportation recommendation with unified route representation learning by exploiting both spatio-temporal dependencies in transportation networks and the semantic coherence of historical routes. Specifically, we first transform the multi-modal transportation network into time-dependent multi-view transportation graphs and devise a graph-based contextual encoder to impute the missing traffic condition in transportation networks by leveraging various contextual factors. Then, we propose a hierarchical multi-task route representation learning ( HMTRL ) framework for recommendations, including (1) a spatiotemporal graph neural network module to capture the spatial and temporal autocorrelation, (2) a coherent-aware attentive route representation learning module to explicitly model route coherence from historical routes, and (3) a hierarchical multi-task learning module to differentiate route representations for different transport modes by incorporating multiple auxiliary tasks equipped in different network layers. Moreover, to improve the model generalization capability, we further propose spatiotemporal pre-training strategies to exploit rich self-supervision signals hidden in transportation networks and historical trajectories. Finally, extensive experimental results on two large-scale real-world datasets demonstrate the effectiveness of the proposed system against eight baselines.

L2RR: Towards a More Diversified Navigation Service with Learning to Rank Routes Framework

Context-aware, Preference-Based Vehicle Routing

Transportation Simulation Modeling and Location-Based Services Data Completion Based on a Data and Model Dual-Driven Approach

Personalized Route Recommendation Based on User Habits for Vehicle Navigation

R4: A Framework for Route Representation and Route Recommendation

Real-time deep reinforcement learning based vehicle navigation

On Representation Learning for Road Networks

Personalized Route Recommendation for Ride-Hailing with Deep Inverse Reinforcement Learning and Real-Time Traffic Conditions

Enhance the performance of navigation: A two-stage machine learning approach

ILRoute: A Graph-based Imitation Learning Method to Unveil Riders’ Routing Strategies in Food Delivery Service

DRL4Route: A Deep Reinforcement Learning Framework for Pick-up and Delivery Route Prediction

Learn-n-Route: Learning implicit preferences for vehicle routing

A Flexible Multi-Layer Map Model Designed for Lane-Level Route Planning in Autonomous Vehicles

Unified route representation learning for multi-modal transportation recommendation with spatiotemporal pre-training

Self-Organized Routing for Autonomous Vehicles Via Deep Reinforcement Learning

Diversified Routing Queries In Dynamic Road Networks

A Reinforcement Learning Framework for Vehicular Network Routing Under Peak and Average Constraints

Transfer Learning-Based Deep Reinforcement Learning Approach for Robust Route Guidance in Mixed Traffic Environment

Book Your Green Wave: Exploiting Navigation Information for Intelligent Traffic Signal Control

Recursive logit-based meta-inverse reinforcement learning for driver-preferred route planning

A Hybrid of Deep Reinforcement Learning and Local Search for the Vehicle Routing Problems