Agentic Framework for Active Data Collection and Travel Behavior Modeling (arXiv 2608.20320)

Paper Overview

• Research area: NLP
• Authors: Narges Ahmadi, Yubo Jiao, Jonatas Augusto Manzolli, Jiangbo Yu, Luis Miranda-Moreno
• Published: 2026-08-22
• arXiv: 2608.20320

Summary

Travel behavior research increasingly combines digital data collection with predictive modeling, yet these stages are often developed and evaluated separately. This study proposes a three-agent workflow integrating conversational data collection, structured data processing, and behavioral prediction.

A chatbot-managed, image-enhanced stated preference survey gathered students' mode choices under five predefined weather scenarios, producing 454 respondent-scenario observations. A processing agent then cleaned and structured the collected data, which a prediction agent used to train weather-sensitive travel demand models.

Key Findings

• The integrated pipeline improved data quality and model interpretability compared to disjointed workflows.
• The agentic architecture supports modular updates and reuse across studies.
• The work offers a blueprint for combining active data collection with behavioral modeling, relevant to adaptive transportation systems and real-time demand management.

Original Abstract (excerpt)

> Travel behavior research increasingly combines digital data collection with predictive modeling, yet these stages are often developed and evaluated separately. This study proposes a three-agent workflow integrating conversational data collection, structured data processing, and behavioral prediction.