Jinhu Qi

Jinhu Qi

Jinhu Qi is a visiting researcher at the HUMANS LAB at the University of Southern California, under the mentorship of Professor Emilio Ferrara. His research centers on Large Language Models (LLMs) applications, particularly in Retrieval-Augmented Generation (RAG) using vector databases and knowledge graphs. Jinhu also explores Chain of Thought (CoT) reasoning, language translation, and misinformation detection. With a Master's in Analytics from USC and a Bachelor's in Computer and Information Science from the University of Oregon, he is currently pursuing a PhD for 2025 to further advance his research in these areas.

Fine-tuning LLMs for Enhanced Tourism Recommendations in Tibet

WorkFlow

Since May 2024, I have joined Professor Wang Ke’s lab at Sichuan University to integrate LLMs with Tibet tourism. One significant challenge we faced was the hallucination problem, where LLMs produced irrelevant or inaccurate responses due to a lack of information about unfamiliar attractions.

To solve this, I fine-tuned LLMs to improve their alignment with tourist needs and local attractions. By collecting descriptions of all tourist attractions in Tibet from the web and fine-tuning the model, I improved hotel information extraction accuracy from 0.47 to 0.98. Comparative experiments using fine-tuning methods like SFT and ORPO resulted in an average score of 80 in accuracy, relevance, and fluency. The fine-tuned models significantly reduced hallucinations compared to baseline models.

Our work was accepted at the ICWOC 2024 conference, and we are now exploring RAG techniques for further optimization. You can find the paper here.