Work

Exploring Search Volumes of Terms in Web Portals for Accurate Event-Aware Traffic Prediction

Year: 2025

Type: conference-paper

Abstract: Traffic prediction is quite challenging in periods of special events (e.g., Thanksgiving and Christmas), where traffic patterns significantly differ from those in normal periods. To address this challenge, we propose leveraging the search volumes of terms available in web online portals as auxiliary data to identify and model such events. After finding that search volumes for traffic-related terms spike during events, we confirm their clear potential for enhancing traffic prediction accuracy by exploring their correlation with traffic flow. Based on these findings, we propose a novel traffic prediction framework, named VESTA, that exploits the search volumes of terms as well as traffic flows. Through extensive experiments, we show VESTA significantly and consistently outperforms state-of- the-art traffic prediction models in accuracy. Our code is available at https://github.com/Bigdasgit/VESTA.

DOI: 10.1145/3701716.3715582

Authors: Dong-Hyuk Seo, Hyomin Shin, Sang-Wook Kim

Institution: Hanyang University

Language: English

Topic: Traffic Prediction and Management Techniques

Subfield: Building and Construction

Field: Engineering

Domain: Physical Sciences

SDG: Sustainable cities and communities

Open Access status: gold

Funders: Universitas Brawijaya, Ministry of Science and ICT, South Korea, Institute for Information and Communications Technology Promotion

Award: G4700831490