Travel Planning with Large Language Models: A Review and Outlook

ACL ARR 2024 December Submission518 Authors

Abstract: Planning is a critical step in advancing artificial intelligence (AI) systems toward higher levels of intelligence and is one of the core capabilities of autonomous decision-making systems, involving complex processes of understanding, reasoning, and decision-making. Current research on planning with AI mostly focuses on simulated environments. Although significant progress has been made, its application in the real world remains limited due to the unpredictable and complex nature of real-world scenarios. Travel planning, as a practical task, is a prime example of these challenges, involving the coordination of factors such as destination selection, budget constraints, and personalized preferences, while also requiring adaptation to changes in external conditions. This review, based on the key roles of LLMs in travel planning tasks, presents a taxonomy of existing methodologies, categorizing them into three types: planner, reformulator, and knowledge source. Furthermore, it outlines directions for future research. We hope this review will provide valuable background information and guidance for researchers in the field, driving the development of this emerging topic.

Paper Type: Long

Research Area: NLP Applications

Research Area Keywords: Large Language Models, Travel Planning, Tourist Trip Design Problem, Natural Language Processing, Agent

Contribution Types: Surveys

Languages Studied: English

Submission Number: 518