What question did this study set out to answer?

This research aims to enhance large language model performance for analyzing social media data in post-occupancy evaluations of underground public spaces.

July 23, 2026ENGINEERING Structure and Civil EngineeringOpen Access

How to optimize large language model performance for human-centric post-occupancy evaluation of underground public space: From model selection to model tuning

View Full Paper

Ask AI

Bookmark

This research aims to enhance large language model performance for analyzing social media data in post-occupancy evaluations of underground public spaces.
Developed a framework for optimizing LLMs through model selection, parameter setting, prompt tuning, and fine-tuning.
Utilized 34 underground public space cases and tested 14 different large language models.
Conducted specialized annotation tasks to assess model performance and accuracy metrics.
Achieved peak macro accuracy of approximately 0.95 and F1-scores over 0.83 for two tasks.
Reasoning models excelled over general-purpose models, achieving F1-score improvements up to 25.74% in complex tasks.
Fine-tuning showed significant performance enhancements and reduced dependency on prompt length.

Authors

CMChen-Xiao MaTongji University FPFang-Le PengTongji University ZLZi-Jian LiTongji University

Discussion

No takes yet. Share an insight, caveat, or question.

Implication

Randomized trial optimizes large language model performance for social media data analysis, suggesting enhanced evaluation of underground public space.

Cite This Study

Ma et al. (2026) studied this question.

synapsesocial.com/papers/6a61af2ffaa9903c5116a1d7 https://doi.org/10.1007/s11709-026-1329-1

View Full Paper

Ask AI

Bookmark

Implication

Randomized trial optimizes large language model performance for social media data analysis, suggesting enhanced evaluation of underground public space.

How to optimize large language model performance for human-centric post-occupancy evaluation of underground public space: From model selection to model tuning

Key Points

Discussion

Implication

Cite This Study

How to optimize large language model performance for human-centric post-occupancy evaluation of underground public space: From model selection to model tuning

Key Points

Discussion

Implication

Cite This Study

Also Consider

Also Consider