AI researcher across speech recognition and GUI agents; middle/last author on 2026 arXiv ASR and navigation work
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) 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 volume cannot be scored yet: the snapshot indexes only 1 OpenAlex work in the last three years (countsByYear: 2026: 1), far below what this axis needs. It would measure total publications and their distribution over time.
Impact—
Impact cannot be scored yet: citation signals (h-index, i10-index, citation counts) are absent or read as zero in this snapshot and are treated as missing, not measured. This axis would capture total citations and influence on follow-up work.
Focus0/100
Focus scores 0: the profile splits evenly (0.5/0.5) between speech processing/ASR and AI work on GUI agents and vision-language models, while OpenAlex topics (Speech Recognition and Synthesis, Speech and Audio Processing, Speech and dialogue systems — 1 work each) show no dominant cluster. Both catalog papers sit in different fields, so concentration is low by construction at this early stage.
Collaboration66/100
Collaboration scores 66, based on 26 distinct coauthors in the record. The pattern is team-based: both 2026 arXiv papers are multi-author efforts with Jian Luan in a middle or last, non-corresponding position, and 12 recurring collaborators — including Yiru Zhang, Hang Su, and Lichun Fan — suggest stable working groups spanning both the speech and GUI-agent lines.
Activity20/100
Activity scores 20: OpenAlex attributes only 1 work in the last three years (recent3yWorks = 1, countsByYear holds a single 2026 entry), even though two 2026 arXiv preprints appear in the Evolve catalog. The indexed record is short, so measured activity looks modest; both known outputs land in 2026, leaving earlier years empty.
Reproducibility20/100
Reproducibility scores 20, the mean over 2 assessed papers (reproMean = 20, reproPapers = 2), both rated low readiness. The GUI-agent MaP paper reports 2-3 point gains on 5 benchmarks without significance tests; CocktailASR-1 rests on proprietary 1M-hour training data with no ablations — strong numbers, limited external replication prospects so far.
Research summary
Jian Luan is an AI researcher whose indexed work spans two areas: speech processing / automatic speech recognition, and artificial intelligence applied to GUI agents and vision-language models — each accounting for half (0.5/0.5) of the field distribution in the OpenAlex-derived snapshot. The profile rests on two 2026 arXiv papers reviewed by Evolve. The first, Xiaomi-CocktailASR-1, is a technical report on LLM-based target-speaker ASR for the cocktail party problem, combining negative sample rejection with chain-of-thought reasoning; Evolve notes strong benchmark numbers but proprietary 1M-hour training data and no ablations or significance testing. The second introduces masked trajectory prediction (MaP) to unify omni-dimensional GUI agent navigation, reporting 2-3 point gains over mixture training across 5 benchmarks, while also lacking significance tests. On both papers Jian Luan appears as a middle or last, non-corresponding author, and the record lists 26 distinct coauthors with 12 recurring collaborators — a team-based pattern. Bibliometric coverage is thin: OpenAlex indexes only 1 work in the last three years (2026), and citation-based indicators are treated as missing rather than zero. Radar scores mirror this: collaboration 66, activity 20, focus 0, reproducibility 20. This is an early-stage profile expected to firm up as indexing catches up.
Collaboration pattern
Jian Luan works in sizeable teams: 26 distinct coauthors appear in the record, with 12 recurring collaborators such as Yiru Zhang, Hang Su, Lichun Fan, Ying Zeng, and Chang Liu. On both 2026 arXiv papers the position is middle or last author rather than corresponding author, consistent with contribution to group technical reports spanning one speech/ASR effort and one GUI-agent effort.
Research keywords
Highlights
LLM-based target-speaker ASR with rejection and chain-of-thought reasoning
Masked trajectory prediction unifies omni-dimensional GUI navigation
A two-field 2026 footprint: speech ASR and GUI agents
The field distribution is an even 0.5/0.5 split between speech processing/ASR and AI for GUI agents/vision-language models; both known papers appeared in 2026, with Jian Luan in middle/last authorship on multi-author teams and 26 distinct coauthors in the record.
Timeline
- 2026Xiaomi-CocktailASR-1 technical report appears on arXiv: LLM-based target-speaker ASR with negative sample rejection and chain-of-thought reasoning (middle authorship, non-corresponding).
- 2026MaP paper on omni-dimensional GUI agent navigation appears on arXiv, reporting 2-3 point gains over mixture training on 5 benchmarks (last authorship, non-corresponding).
- 2026OpenAlex snapshot records 1 indexed work for 2026 — the only year with activity in the current profile, with citation-based indicators still missing.
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.
- Bibliometrics are OpenAlex-derived (identity: openalex); h-index and i10-index read 0 in the snapshot and are treated as missing rather than true measures.
- Institution, country, and ORCID are absent from the input data; this profile is name-resolved from public bibliographic data and homonyms cannot be ruled out.
- Only two 2026 arXiv preprints were assessed by Evolve and both rate low on reproducibility readiness, so field, focus, and reproducibility scores rest on a very small base.
- Neither catalog paper reports significance tests, and CocktailASR-1 relies on proprietary 1M-hour data; benchmark claims are relayed from the review verdicts, not independently verified.