Travel agents compared for fixing disrupted itineraries effectively and efficiently
Replan, Repair, or Edit? A Unified Empirical Evaluation of Travel Agents for Itinerary Revision under Resource Disruptions
Artificial IntelligenceMultiagent Systems
Summary
Sometimes travel plans like flights or hotels get canceled, making your original itinerary impossible to follow. This paper studies different methods that can automatically fix such travel plans after problems occur. The authors test full replanning, step-by-step repair, and AI-powered local changes to see which works best across various scenarios. They find that full replanning handles complex issues well, but step-by-step repairs keep more of the original plan and require fewer changes. They also compare how much computing each approach needs, offering advice on when to use each method.
What this means in practice
- •For travel technology developers: Improve travel booking apps by selecting the best automated approach to fix user itineraries after disruptions like cancellations.
- •For logistics planners: Adapt complex schedules with minimal changes when unexpected resource disruptions occur by using repair or revision techniques tested in this study.
Authors
Xiaofei Yuan, Yan Zhang, Shaobo Qiao, Huangleshuai He, Leyan Ni, Mingchen Ju, Lujia Yang, Sijia Xu, Yifu Tang, Zhengyi Yang
Abstract
Travel-planning agents generate itineraries that may become infeasible after acceptance because of flight cancellations, hotel unavailability, or attraction closures. Revising these itineraries involves full replanning, classical plan repair, and LLM-based travel-agent revision, whose differing task formulations and evaluation protocols hinder comparison. We conduct a systematic empirical study using two TREK-derived benchmark sets: 500 single-disruption cases, including feasible and infeasible instances, and 200 feasible simultaneous compound-disruption cases. We compare LLM-Z3 full replanning, IPyHOPPER hierarchical repair, and an iTIMO local-revision adapter across effectiveness, plan stability, and computational cost. LLM-Z3 with Gemini achieved the highest observed compound-disruption success. IPyHOPPER nearly matched that configuration's single-disruption overall success, while preserving substantially more of the accepted itinerary on successful repairs. Successful hierarchical and local repairs made fewer edits and retained more accepted commitments than full replanning. Computational profiles differed: IPyHOPPER used no LLM inference, the evaluated LLM-Z3 adapter used compact one-call inference, and the iTIMO adapter consumed substantially more tokens. The study provides practical guidelines for balancing feasibility recovery, commitment preservation, and computational cost within evaluated settings.