Fang-Yu Hsu

Master's

Computer Science

SkimLit

- Developed an NLP model to summarize academic abstracts into specific categories (e.g., objective, methods, results), enabling faster comprehension of key components - Constructed deep architectures using pretrained token, character, and positional embeddings for higher accuracy - Outperformed a TF-IDF baseline model, achieving 85% accuracy in predictions with 257M+ parameters - Automated RCT abstract testing from PubMed, generating summarization txt reports to streamline reviews