National Taiwan University NLP researcher: LLM-driven biomedical text simplification and continual LLM adaptation
2
papers here
2
reviewed by Evolve
2
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Output is not yet scoreable: beyond the two 2026 arXiv preprints in the Evolve catalog, no publication counts are available, and OpenAlex records an h-index of 0 with no broader corpus to measure volume.
Impact—
No citation data is available yet: h-index and i10-index are recorded as 0 and both works are recent 2026 preprints, so this axis cannot measure citation influence until citations accumulate.
Focus0/100
Focus scores 0, driven directly by the field distribution: it splits exactly 0.5/0.5 between natural language processing for biomedical text simplification and for LLM continual adaptation, and the topic list (Topic Modeling and NLP techniques at 2 mentions each; domain adaptation/few-shot learning and text readability at 1 each) confirms two parallel lines rather than one dominant specialty.
Collaboration39/100
The collaboration score is 39, built on a small network: OpenAlex records 6 distinct co-authors across the cataloged work. Both first-authored papers are co-written within one institutional group — frequent collaborators include Hsin-Hsi Chen, Hen-Hsen Huang, and Po-Chun Chen — so the pattern is a stable, mentor-supported team rather than wide external reach. The score reflects this narrow but consistent circle.
Activity32/100
Activity scores 32 on a very short record: OpenAlex countsByYear shows a single entry, 2026 with 2 works, and recent3yWorks is likewise 2 — the entire visible output falls in one year. This makes the pace look modest next to established profiles, but the measure covers only these two first-authored arXiv preprints, so the axis should be read as early-career, not dormant.
Reproducibility38/100
Reproducibility averages 38 across the 2 papers with Evolve reproducibility assessments. The PLABA-2024 medical-simplification study is rated low: Evolve notes the mixture-of-agents gains are unsupported and statistical tests absent. The continual-adaptation paper is rated medium, with real but task-dependent gains (29/42 versus Self-StreamICL) shown under a single seed. Consistent statistical reporting would lift this axis most directly.
Research summary
Ting-Wei Chang is a natural language processing researcher at National Taiwan University whose cataloged work centers on two complementary problems: making biomedical text accessible to non-expert readers, and enabling large language models to keep improving from live environment feedback. Both papers in the Evolve catalog are first-authored arXiv preprints from 2026. The first, a broad PLABA-2024 study, benchmarks prompting, QLoRA fine-tuning, and mixture-of-agents pipelines for plain-language adaptation of medical texts; Evolve's review finds the empirical sweep solid but flags that mixture-of-agents gains are unsupported and statistical testing absent (reproducibility readiness: low). The second proposes environment-driven dynamic policies for continual LLM improvement, showing real but task-dependent gains (29/42 versus Self-StreamICL) under a single-seed evaluation (readiness: medium). OpenAlex-based bibliometrics reflect an early-stage record: two works in 2026, h-index and i10-index each recorded as 0, six distinct co-authors, and a field distribution split evenly (0.5/0.5) between biomedical text simplification and LLM continual adaptation. Radar scores place collaboration at 39, activity at 32, and reproducibility at 38, with focus scored 0 given the even two-field split; output and impact axes lack sufficient data. Frequent co-authors include Hsin-Hsi Chen, Hen-Hsen Huang, and Po-Chun Chen.
Collaboration pattern
Ting-Wei Chang publishes as first author within a compact, mentor-supported team: 6 distinct co-authors appear across the cataloged work, with Hsin-Hsi Chen and Hen-Hsen Huang recurring on both papers, plus Po-Chun Chen. Both 2026 preprints are first-authored without corresponding-author credit, consistent with an early-career researcher publishing alongside senior NLP co-authors at National Taiwan University. No external or cross-institution collaboration appears in the available data.
Research keywords
Highlights
Plain-language adaptation of medical texts with LLMs (PLABA-2024)
Environment-driven dynamic policies for continual LLM improvement
Timeline
- 2026First-authored arXiv preprint on LLM-driven plain language adaptation of medical texts, a broad PLABA-2024 study of prompting, QLoRA, and mixture-of-agents.
- 2026First-authored arXiv preprint proposing environment-driven dynamic policies for continual LLM improvement in streaming settings.
- 2026OpenAlex snapshot consolidates around this year: 2 cataloged works, 6 distinct co-authors, and h-index and i10-index both recorded as 0.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric indicators (radar scores, co-author counts, h-index, i10-index, counts by year) come from OpenAlex; identity is name-resolved from public bibliographic data, with no ORCID linked.
- Only 2 first-authored 2026 arXiv preprints are in the Evolve catalog; this snapshot captures no peer-reviewed venue publications, grants, or awards.
- Output and impact radar axes are unscored because citation and volume data are insufficient (h-index and i10-index recorded as 0).
- Evolve reproducibility assessments cover both papers; review verdicts reflect Evolve's editorial judgment, not external peer review.