Hoàng
Building biomedical NLP systems

Research systems · 2026

Projects, as working arguments.

Each case study starts with a question and keeps the unfinished parts visible: constraints, system boundaries, failed approaches, and the next experiment that could change the design.

Active records
03
Reporting rule
No unlabelled metrics
P–01Primary investigation

Clinical Entity Recognition and Linking Pipeline

A Vietnamese clinical NLP pipeline that separates span detection, assertion handling, candidate retrieval, and ontology linking so each source of error can be inspected independently.

Active research and competition system
Entity path · deterministic interfaces
01Note textnormalized tokens
02XLM-R + CRFtyped spans
03Assertioncontext state
04ALexicalalias overlap
04BDensesemantic context
05Rerank + linkRxNorm / ICD
Fig. 01Recognition and assertion stay observable before two retrieval paths merge.
Research question

How can a Vietnamese clinical system preserve precise entity boundaries and assertion context while linking noisy surface forms to controlled medical vocabularies with limited annotated data?

  • XLM-R
  • Linear-chain CRF
  • PyTorch
  • BM25
  • Dense retrieval
  • Cross-encoder reranking
Open system case study
P–02Experimental pipeline

POV Video Segmentation and Keyframe Selection

A shot-aware preprocessing pipeline for long first-person video: locate visual boundaries, sample representative frames, embed them, and remove near-duplicates without erasing meaningful transitions.

Experimental pipeline
  • TransNetV2
  • FFmpeg
  • SigLIP2
  • Python
  • Cosine similarity
Inspect the temporal pipeline
Temporal trace · boundary-aware sampling
Shot segmentation and keyframe selection timelineA video is divided at two inferred shot boundaries. Candidate frames are embedded, and three representative frames remain after local similarity filtering.SHOT 01SHOT 02SHOT 03keyframe · 01keyframe · 02keyframe · 03
Fig. 02Large points survive local cosine filtering; shot identity and timestamps remain attached.
Learning system

AI Algorithms Learning Laboratory

A structured learning program that moves from mathematical derivation to implementation, tiny verification cases, and visual explanation while preparing for the 2026 Vietnamese Student AI Olympiad.

Which small implementations and visual tests reveal the assumptions of an AI algorithm more clearly than using a high-level library alone?
01 / Derive02 / Implement03 / Interrogate04 / Verify
Ongoing learning program
Read the working log