Peng Cheng Laboratory researcher spanning structured-data foundation models and lifelong conversational agents
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
0
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-14) 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—
Publication-output metrics are not yet available for this profile: no total-work counts beyond the recent window, and the recorded h-index and i10-index are 0. This axis would measure cumulative and per-year publication volume, so no score is given.
Impact—
Citation-based impact cannot be scored: the source records h-index and i10-index as 0 with no citation counts attached. This axis would measure citation accumulation, field-normalized influence, and uptake of the reviewed preprints, none of which is yet observable.
Focus0/100
Focus scores 0/100, reflecting maximal thematic dispersion in the available data: the two reviewed 2026 papers sit in different fields (machine learning for tabular data and conversational NLP, split 0.5/0.5), and OpenAlex assigns five topics with one work each — geometric analysis, logic and reasoning, Bayesian causal inference, intelligent tutoring, and explainable AI. On this evidence the profile reads as exploratory rather than concentrated, and the score may consolidate as a core research line emerges.
Collaboration84/100
The strongest dimension at 84/100, driven by 67 distinct coauthors recorded in OpenAlex against only 3 recent works — an unusually large collaboration network. On the two reviewed papers Wang appears as a middle-position, non-corresponding author, and named frequent collaborators include Xingxuan Zhang, Gang Ren, Hao Yuan, and Jinghan Zhang. The pattern is consistent with intensive team-based research within large laboratory groups rather than solo or small-core work.
Activity40/100
Moderate at 40/100. The countsByYear record shows 3 works in 2026, and recent3yWorks is likewise 3, meaning all recorded output is concentrated in the current year with no longer publication history visible in this data source. This is consistent with an early-stage or newly attributed profile whose activity level will need re-scoring as further works accrue.
Reproducibility38/100
Scored 38/100 from Evolve's reproducibility assessment of 2 papers (reproMean 38). LimiX-2 is rated medium readiness — broad benchmark evidence, but overlapping top-rank Elo confidence intervals and one self-published benchmark — while ThinkFlow is rated low, with duplicated ablation rows and closed-source overclaims cited. Releasing code and data and tightening ablation reporting would most directly move this axis.
Research summary
Hui Wang is a researcher at Peng Cheng Laboratory (China) whose profile is resolved from OpenAlex public bibliographic data. The recorded output is recent and concentrated: 3 works in 2026, all falling within the last three years. The Evolve-reviewed catalog covers two 2026 arXiv preprints — LimiX-2 on general structured-data intelligence and ThinkFlow on lifelong conversational agents — in both of which Wang is a middle-position, non-corresponding author, consistent with a contributor role inside larger teams. The recorded research footprint splits evenly (0.5/0.5) between machine learning for tabular data and conversational NLP, while OpenAlex topic tags spread across five single-count topics, from geometric analysis to explainable AI, indicating unusually wide thematic reach for a young profile. Collaboration is the standout signal: 67 distinct coauthors are recorded, yielding a high collaboration score of 84/100. Reproducibility is the weaker axis, with a mean readiness of 38 across 2 assessed papers and review notes citing duplicated ablations and closed-source claims. Citation-based output and impact metrics are not yet available, so this profile should be read as early-stage.
Collaboration pattern
The collaboration signal is strong: OpenAlex records 67 distinct coauthors against just 3 recent works, and the dimension scores 84/100. On both reviewed 2026 papers Wang is a middle-position, non-corresponding author, indicating work within sizeable teams. Named frequent collaborators include Xingxuan Zhang, Gang Ren, Hao Yuan, Hongze Tan, and Jinghan Zhang, among others — a pattern typical of project-based collaboration inside large laboratory groups rather than a stable two-person partnership.
Research keywords
Highlights
LimiX-2: pushing tabular in-context learning toward general structured-data intelligence
ThinkFlow: self-evolving latent memory for lifelong conversational agents
Timeline
- 2026LimiX-2 preprint on a Contextual Mechanism Network for general structured-data intelligence released on arXiv (middle author, not corresponding).
- 2026ThinkFlow preprint on self-evolving probabilistic latent memory for lifelong conversational agents released on arXiv (middle author, not corresponding).
- 2026OpenAlex records 3 works for this year, all falling within the recent three-year window; 67 distinct coauthors are recorded in total.
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 data comes from OpenAlex only; no ORCID is linked, and 'Hui Wang' is a common name, so this profile is name-resolved from public bibliographic records.
- Output and impact dimensions are unscored because citation data is absent (h-index and i10-index recorded as 0), not because impact was measured as zero.
- The Evolve-reviewed sample is small: 2 arXiv preprints, both from 2026, with middle non-corresponding authorship, so verdicts may not represent the full research portfolio.
- countsByYear covers only 2026, so no multi-year activity or productivity trend can be established.