Yun Liu

Tsinghua University

Unclaimed · AI-structured profileOpenAlex

Yun Liu (Tsinghua University): NLP researcher across legal-AI reasoning and Japanese speech recognition

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-02) 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—

Underlying data is insufficient: no publication counts, venue list, or citation-linked output metrics are available in this snapshot. This axis would measure total research output and its distribution over venues and years; it is left unscored rather than estimated.

Impact—

No citation counts, h-index, or downstream-uptake signals are present for this profile, so impact cannot be scored. This axis would measure how widely the work is cited and used by other researchers; it remains unmeasured here.

Focus0/100

The radar focus score is 0, the lowest value on the 0-100 scale, and the raw field distribution explains it: the two reviewed papers split exactly evenly (0.5/0.5) between legal-AI NLP and automatic speech recognition. With only two cataloged papers, this even split may reflect a small sample rather than a settled agenda; more papers would be needed to tell whether the strands converge or one dominates.

Collaboration58/100

The radar collaboration score is 58, backed by 17 distinct coauthors, of whom 12 recur as frequent collaborators (including Qingjing Chen, Junkai Zhang, Yukun Yan, and Jun Liu). Both cataloged papers place Liu as a middle author, a pattern consistent with contribution inside larger multi-author teams spanning two subfields. The moderate score and the even two-field split suggest an active but not yet central collaboration footprint in this snapshot.

Activity—

Recent publication frequency and year-over-year continuity cannot be computed from this snapshot, which contains only two 2026 papers. This axis would measure ongoing research tempo and sustained output over time; it is left unscored.

Reproducibility55/100

The radar reproducibility mean is 55, computed over 2 assessed papers, and both carry a medium repro-readiness rating. Evidence is usable but not sealed: Ruby-ASR reports gains on five benchmarks with code released, yet reading-label provenance and speech causality are unverified; LEGO shows strong same-backbone RAG gains, but its parity claim with larger models lacks statistical support.

Research summary

Yun Liu is a researcher at Tsinghua University whose reviewed work spans two strands of natural-language technology: legal AI and automatic speech recognition. Evolve's catalog includes two 2026 arXiv papers with Liu as a middle, non-corresponding author. The first, LEGO, combines expert-graph retrieval-augmented generation with syllogistic chain-of-thought for legal reasoning over the Chinese Civil Code, reporting strong gains over same-backbone RAG baselines, though its claimed parity with larger models lacks statistical support. The second, Ruby-ASR, introduces evidence-preserving ruby span-bound supervision for Japanese ASR that jointly handles orthographic transcription and lexical reading, reporting gains across five benchmarks with code released, while reading-label provenance and speech causality remain unverified. Bibliometric radar data records 17 distinct coauthors, 12 of whom recur frequently, a collaboration score of 58, and a reproducibility mean of 55 across the two assessed papers; the field distribution splits evenly (0.5/0.5) between the two strands. Output, impact, and activity metrics are unavailable in this snapshot. Because 'Yun Liu' is a common name, this profile is name-resolved from public bibliographic data.

Collaboration pattern

Liu's collaboration footprint in this snapshot covers 17 distinct coauthors, 12 of them recurring — including Qingjing Chen, Junkai Zhang, Shaochun Wang, Yukun Yan, and Jun Liu. A collaboration score of 58 with middle-author placement on both papers points to steady participation in multi-author teams rather than solo or leading roles, and the pool spans both subfields, matching the even 0.5/0.5 field split.

Research keywords

Japanese ASRruby/furigana annotationlexical reading disambiguationkanji polyphonygrapheme-to-phoneme conversionCTCGraphRAGlegal reasoningchain-of-thoughtretrieval-augmented generationknowledge graphChinese Civil Code

Highlights

LEGO: expert-graph RAG plus syllogistic chain-of-thought for Chinese civil-law reasoning
In this 2026 arXiv paper Liu contributes as a middle author to a system pairing an expert knowledge graph with retrieval-augmented generation and a syllogistic chain-of-thought pipeline for legal reasoning over the Chinese Civil Code, reporting strong gains over same-backbone RAG baselines; Evolve notes the claimed parity with larger models lacks statistical support.
Ruby-ASR: evidence-preserving supervision for Japanese lexical-reading recognition
Liu is a middle author on this 2026 arXiv paper introducing ruby span-bound targets that jointly supervise orthographic transcription and lexical reading for Japanese ASR, disambiguating kanji polyphony; it reports gains across five benchmarks and releases code, while reading-label provenance and speech causality remain unverified.
A dual-track language-technology agenda within one year
The two reviewed papers divide exactly evenly between legal-AI NLP and automatic speech recognition (field distribution 0.5/0.5), and both appeared as 2026 arXiv preprints with middle authorship, suggesting Liu participates in two distinct language-technology threads at Tsinghua rather than a single specialized track.

Timeline

2026LEGO, combining expert-graph retrieval-augmented generation with syllogistic chain-of-thought for Chinese civil-law legal reasoning, appears on arXiv with Liu as a middle author.
2026Ruby-ASR introduces evidence-preserving ruby span-bound supervision for joint orthographic and lexical-reading Japanese ASR, reporting five-benchmark gains with released code.
2026Both cataloged papers land in the same year across the two halves of the field distribution (0.5/0.5), within a coauthor network of 17 distinct collaborators.

Papers in the Evolve catalog

Frequent co-authors

Qingjing Chen×1 Junkai Zhang×1 Shaochun Wang×1 Jiahao Ding×1 Siyuan Zheng×1 Yukun Yan×1 Zhi Zheng×1 Antonino Rotolo×1 Weixing Shen×1 Hao Shi×1 Xuehao Yang×1 Jun Liu×1

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.

Radar scores and counts come from a limited bibliometric snapshot; only two reviewed papers (both 2026 arXiv preprints, middle authorship) underlie the field, collaboration, and reproducibility axes.
Output, impact, and activity metrics (publication counts, citations, yearly tempo) are unavailable in this snapshot and left unscored.
'Yun Liu' is a common name and no ORCID is linked, so this profile is name-resolved from public bibliographic data and may mix records of different individuals.
Affiliation country, career history, and peer-review status of the preprints are not provided; statements are limited to the reviewed papers and radar data.