Ruoming Jin

Kent State University · United States

Unclaimed · AI-structured profileOpenAlex

Kent State researcher working across linear recommender theory and parameter-efficient personalization of LLMs

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

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—

Not scored: the snapshot provides no reliable publication-volume or citation data; h-index and i10-index are recorded as 0, which reflects missing author-profile coverage rather than measured output. This axis would measure cumulative publications, venue quality, and productivity trend over time.

Impact—

Not scored: citation counts, citation trajectory, and h-index-based influence data are unavailable in the current snapshot. This axis would measure how widely the work is cited and its influence within recommender systems and LLM personalization research.

Focus0/100

The focus score is 0, reflecting an unusually even spread: the field distribution splits into three equal thirds — parameter-efficient LLM adaptation, recommender systems, and LLM personalization — and OpenAlex topics are similarly dispersed (Topic Modeling: 2; Recommender Systems and Techniques: 2; Domain Adaptation and Few-Shot Learning: 2; Natural Language Processing Techniques: 1; Cryptography and Data Security: 1). The threads are linked by low-rank modeling and personalization rather than one narrow specialty.

Collaboration54/100

Collaboration scores 54. The recent record involves 14 distinct coauthors, with a recurring cluster including Xinyu Li, Jianfeng Zhu, Ruixin Guo, Zhi Liu, Hao Zhou, Dong Li, Zhenming Liu, Jing Gao, Bin Ren and Julina Maharjan. Across the three reviewed papers the author position is middle or last, and none is a corresponding-author paper, suggesting a contributing rather than leading role in these teams.

Activity47/100

Activity scores 47. The snapshot records 4 works in the last three years, and the counts-by-year data contains only 2026 (4 works), so visible output is concentrated in a single recent year of arXiv preprints. With no earlier-year counts provided, the trajectory before 2026 cannot be assessed; the moderate score reflects thin but current recent volume.

Reproducibility55/100

Reproducibility averages 55 across the 3 papers Evolve has reviewed, each rated medium on repro readiness. The linear-recommender paper reports closed forms matching ADMM but no variance; APLAUD and PLUME show solid storage or parameter savings yet thin accuracy gains without significance testing. Methods are typically well specified, but statistical rigor is the recurring gap.

Research summary

Ruoming Jin is a researcher at Kent State University whose recent work, as name-resolved from OpenAlex bibliographic data, sits at the intersection of recommender systems and parameter-efficient adaptation of large language models. The Evolve catalog contains three 2026 arXiv papers with this author, and the field distribution splits into three equal thirds: machine learning/NLP (parameter-efficient LLM adaptation), recommender systems, and NLP (LLM personalization). One thread is theoretical: "On the Regularization Landscape for the Linear Recommendation Models" unifies linear recommenders as nuclear- versus Frobenius-norm regularization, with closed forms matching ADMM, though Evolve notes thin empirical gains and no variance reporting. The other thread is personalization: APLAUD and PLUME both build shared-subspace LoRA methods for per-user LLMs, with PLUME — where the author is last author — claiming OPPU-level personalization from just 3-7% of per-user LoRA parameters. Evolve's verdicts converge on the same caution: storage or parameter savings are solid, but accuracy gains are small and lack significance testing. Bibliometric signals are thin: 4 works in the last three years, 14 distinct coauthors, an h-index recorded as 0 (indicating missing coverage), and a reproducibility mean of 55 across three medium-readiness papers. Collaboration (54) and activity (47) are mid-range; the 0 focus score reflects an even three-way spread of topics rather than narrow specialization.

Collaboration pattern

The collaboration score is 54, mid-range. Recent work involves 14 distinct coauthors, with a recurring group of Xinyu Li, Jianfeng Zhu, Ruixin Guo, Zhi Liu, Hao Zhou, Dong Li, Zhenming Liu, Jing Gao, Bin Ren and Julina Maharjan. Across the three reviewed papers the author sits in a middle or last position and is never the corresponding author, consistent with a contributing role in small-to-mid-sized machine learning teams rather than leading them.

Research keywords

matrix factorizationcollaborative filteringnuclear norm regularizationFrobenius norm regularizationlinear autoencodersLLM personalizationparameter-efficient fine-tuningLoRAlow-rank adaptationshared subspaceper-user adapters

Highlights

Unifying linear recommenders via nuclear- vs Frobenius-norm regularization
The 2026 arXiv paper On the Regularization Landscape for the Linear Recommendation Models casts linear collaborative filtering, including EASE-style models, within a single regularization landscape with closed forms matching ADMM; Evolve finds the theory neat but the empirical gains thin and no variance reported.
APLAUD: adaptive low-rank decomposition for user-specific LLMs
This 2026 arXiv paper proposes an adaptive personalized low-rank decomposition in a shared subspace for per-user survey prediction; Evolve judges the storage savings solid but the small accuracy gains over OPPU untested for statistical significance, with medium repro readiness.
PLUME: OPPU-level personalization from 3-7% of per-user LoRA parameters
As last author on this 2026 arXiv paper, the researcher contributes a low-rank user-modulation method in shared subspaces; Evolve finds the claim plausible but the performance gains marginal, significance testing absent, and one baseline result excluded.

Timeline

2026Published "On the Regularization Landscape for the Linear Recommendation Models" on arXiv, unifying linear recommenders under nuclear- versus Frobenius-norm regularization.
2026Published APLAUD on arXiv, an adaptive personalized low-rank decomposition method for user-specific large language models.
2026Published PLUME as last author, claiming OPPU-level LLM personalization with only 3-7% of per-user LoRA parameters.
2026Bibliometric snapshot records 4 works in the last three years, all dated 2026, involving 14 distinct coauthors.

Papers in the Evolve catalog

Frequent co-authors

Xinyu Li×2 Jianfeng Zhu×2 Ruixin Guo×1 Zhi Liu×1 Dong Li×1 Zhenming Liu×1 Hao Zhou×1 Zhi Liu×1 Jing Gao×1 Bin Ren×1 Hao Zhou×1 Julina Maharjan×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.

Bibliometrics come from OpenAlex only; the snapshot records h-index and i10-index as 0, which indicates incomplete author-profile coverage rather than measured output.
The output and impact dimensions are unscored because publication-volume and citation data are unavailable in this snapshot.
This profile is name-resolved from public bibliographic data (identity: openalex); no ORCID is linked, so attribution to a single individual should be verified before citing.
Catalog coverage is limited to three 2026 arXiv preprints reviewed by Evolve; earlier career history, venues, and funding fall outside this data scope.