Resume

M.S. in Data Science, National Taiwan University Sep. 2021 - Aug. 2023

B.S. in Engineering and System Science, National Tsing Hua University Sep. 2017 - Jan. 2021

FincGAN: A Gan Framework of Imbalanced Node Classification on Heterogeneous Graph Neural Network

Hung-Chun Hsu*, Ting-Le Lin*, Bo-Jun Wu, Ming-Yi Hong, Che Lin, Chih-Yu Wang

In Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024

FlashGAN: Framework of Localized Node Augmentation via Semi-supervised Learning in Heterogeneous Graphs with Generative Adversarial Network

Hung-Chun Hsu, Bo-Jun Wu, Ming-Yi Hong, Che Lin, Chih-Yu Wang

arXiv preprint arXiv:2312.06519, 2024

Test-Time Scaling Strategies for Generative Retrieval in Multimodal Conversational Recommendations

Hung-Chun Hsu, Yuan-Ching Kuo, Chao-Han Huck Yang, Szu-Wei Fu, Hanrong Ye, Hongxu Yin, Yu-Chiang Frank Wang, Ming-Feng Tsai, Chuan-Ju Wang

arXiv preprint arXiv:2508.18132, 2025

CFDA & CLIP at TREC iKAT 2025: Enhancing Personalized Conversational Search via Query Reformulation and Rank Fusion

Yu-Cheng Chang, Guan-Wei Yeo, Quah Eugene, Fan-Jie Shih, Yuan-Ching Kuo, Tsung-En Yu, Hung-Chun Hsu, Ming-Feng Tsai, Chuan-Ju Wang

NIST Text REtrieval Conference (TREC), 2025

LODESTAR: Trustworthy Entropy Is Navigated, Not Merely Measured

Hung-Chun Hsu, Po-Jen Ko, Che-Cheng Wu, Li-Yang Chang, Chuan-Ju Wang

arXiv preprint arXiv:2608.11922, 2026

Senior Research Assistant and Project Lead, CFDA Lab, Academia Sinica Apr. 2024 - Present

• Leading the Information Retrieval (IR) research project in collaboration with NVIDIA Research (Taiwan and US) under the supervision of Prof. Chuan-Ju Wang and Prof. Ming-Feng Tsai. Focused on conversational retrieval and Multimodal-LLMs.
• Participated in NIST's Text Retrieval Conference (TREC) tracks since 2024: Product Recommendations, Interactive Knowledge Assistance (iKAT), and User Simulation. Led and coordinated the competition teams of up to 7 members.

Data Scientist Intern - R&D, Cathay Financial Holdings Feb. 2024 - Apr. 2024

• Developed an enterprise-level Chinese legal retrieval pipeline leveraging LangChain, increasing recall@50 performance from 23% to 39% compared to LangChain's basic two-staged retrieval settings through a novel legal articles pre-classification approach.

Research Assistant, SNAC Lab, Academia Sinica Sep. 2021 - Feb. 2024
Full-time (2023 - 2024) and Part-time (2021 - 2023)

• Developed novel GAN-based data augmentation frameworks for heterogeneous graphs, addressing critical node class imbalance issues in graph neural networks.
• Achieved 14.4% improvement in F-score and 12.3% in PR-AUC over SOTA baseline GraphSMOTE through innovative handling of structured graph G=(V,E,X). Published research in IEEE ICASSP with ongoing follow-up studies.

2025 TREC Interactive Knowledge Assistance Track (iKAT) Jun. 2025 - Sep. 2025

• Led a team of 7 members to participate in the iKAT track, developing pipelines for traditional offline conversational search and exploring real-time conversational retrieval methods requiring low latency.
• Achieved 2nd place in Offline Passage Ranking and 3rd place in Online Interactive Generation among participating teams, outperforming the competition median by 10 pts in nDCG@3 and 11 pts in average human evaluation score respectively.

Multimodal Conversational Product Search with MLLMs, with NVIDIA Research Taiwan Apr. 2024 - Jul. 2025

• Developed a product recommendation system supporting any-to-any multimodal data using a novel MLLM-based generative retrieval to predict consumer purchases, enhanced by our test-time reranking mechanism with average improvements of 14.5 points MRR and 10.6 points nDCG@1 across 3 benchmarks.

2024 TREC Product Search and Recommendations Track Jun. 2024 - Sep. 2024

• Led a 4-member team in the product search and recommendations track. Developed a multimodal product retrieval system leveraging BLIP, ViLT, and BEiT-3 as dual encoders and cross encoders.
• Implemented Weighted Sum Fusion to combine unimodal and multimodal retrieval results, achieving nDCG@10 of 76% compared to 72% from traditional BM25 and SPLADE retrievers.

Insurance Claim Fraud Prediction, Cathay Life Insurance Co., Ltd. & National Taiwan University Sep. 2021 - Sep. 2022

• Designed and implemented heterogeneous social network graphs for large-scale networks and developed data augmentation framework to enhance graph neural network performance in detecting hard-to-identify minority fraud risk accounts.

Scholastic Honor Membership 2023

awarded by The Phi-Tau-Phi Scholastic Honor Society of the R.O.C. (Taiwan)

Graduation Ceremony Representative 2023

awarded by Data Science Master Program, National Taiwan University and Academia Sinica

Academic Excellence Scholarship 2023

awarded by National Taiwan University and Academia Sinica

2nd Place in National Data Competition, Carrefour Taiwan E-commerce Next Purchase Prediction 2022

awarded by Institute for Information Industry, Ministry of Digital Affairs, Executive Yuan of the R.O.C. (Taiwan)