Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.03308 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915979298078720 |
|---|---|
| author | Zhang, Bo-Wen Ye, Jin Hua, Peng-Yu Cao, Jia-Wei Shao, Jie-Jing Li, Yu-Feng Guo, Lan-Zhe |
| author_facet | Zhang, Bo-Wen Ye, Jin Hua, Peng-Yu Cao, Jia-Wei Shao, Jie-Jing Li, Yu-Feng Guo, Lan-Zhe |
| contents | Travel planning serves as a critical task for long-horizon reasoning, exposing significant deficits in LLMs. However, existing benchmarks and evaluations primarily assess final plans in an end-to-end manner, which lacks interpretability and makes it difficult to analyze the root causes of failures. To bridge this gap, we decompose travel planning into five constituent atomic sub-capabilities, including \emph{Constraint Extraction}, \emph{Tool Use}, \emph{Plan Generation}, \emph{Error Identification}, and \emph{Error Correction}. We implement a decoupled evaluation protocol leveraging oracle intermediate contexts to rigorously isolate these components, thereby measuring the atomic performance boundary without the noise of cascading errors. Our results highlight a clear contrast in performance: while LLMs are proficient in extracting explicit constraints, they struggle to infer implicit, open-world requirements. Furthermore, they exhibit structural biases in plan generation and suffer from ineffective self-correction, characterized by excessive sensitivity and erroneous persistence. These findings offer precise directions for improving LLM reasoning and planning abilities. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_03308 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Revisiting the Travel Planning Capabilities of Large Language Models Zhang, Bo-Wen Ye, Jin Hua, Peng-Yu Cao, Jia-Wei Shao, Jie-Jing Li, Yu-Feng Guo, Lan-Zhe Artificial Intelligence Travel planning serves as a critical task for long-horizon reasoning, exposing significant deficits in LLMs. However, existing benchmarks and evaluations primarily assess final plans in an end-to-end manner, which lacks interpretability and makes it difficult to analyze the root causes of failures. To bridge this gap, we decompose travel planning into five constituent atomic sub-capabilities, including \emph{Constraint Extraction}, \emph{Tool Use}, \emph{Plan Generation}, \emph{Error Identification}, and \emph{Error Correction}. We implement a decoupled evaluation protocol leveraging oracle intermediate contexts to rigorously isolate these components, thereby measuring the atomic performance boundary without the noise of cascading errors. Our results highlight a clear contrast in performance: while LLMs are proficient in extracting explicit constraints, they struggle to infer implicit, open-world requirements. Furthermore, they exhibit structural biases in plan generation and suffer from ineffective self-correction, characterized by excessive sensitivity and erroneous persistence. These findings offer precise directions for improving LLM reasoning and planning abilities. |
| title | Revisiting the Travel Planning Capabilities of Large Language Models |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2605.03308 |