Jay Pujara

University of Southern California · United States

Unclaimed · AI-structured profileOpenAlex

USC researcher building knowledge graphs from scientific publications and representation learning for uncertain KGs.

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—

No output score is available yet: publication-volume metadata for this OpenAlex identity is incomplete. This axis would measure total works over time, venue spread, and authorship roles, beyond the seven recent works already counted.

Impact—

Impact is unscored: citation counts and citing-work data are not available for this identity, whose h-index and i10-index are recorded as 0. This axis would measure citation influence and uptake of the reviewed methods across the literature.

Focus0/100

Focus scores 0 in this snapshot, driven by a field distribution split evenly at 0.5 between machine learning for knowledge graph representation learning and knowledge graph construction with scientific information extraction. Topic counts reinforce breadth over concentration: topic modeling (4), semantic web and ontologies (2), advanced graph neural networks (2), scientometrics (1), and NLP techniques (1). Both strands still converge on knowledge graphs as the shared object of study.

Collaboration32/100

Collaboration scores 32, based on 4 distinct coauthors recorded in this OpenAlex snapshot — a compact network. Recurring partners include Craig A. Knoblock, Md Abrar Jahin, and Taufikur Rahman Fuad. On the two Evolve-reviewed 2026 papers Pujara appears as last author on ArticleMiner and a middle author on the spectral initialization paper, and is not marked corresponding on either, suggesting senior-supervision and contributor roles within a small team.

Activity61/100

Activity scores 61. The snapshot records 7 works in the last three years, with year counts of 1 in 2024 and 6 in 2026 — output is heavily concentrated in 2026, including both Evolve-reviewed arXiv papers. This suggests an active but recently re-accelerating publishing cadence rather than steady annual volume; earlier-year counts are absent from this snapshot.

Reproducibility55/100

Reproducibility scores 55, computed as a mean of 55 across the 2 Evolve-reviewed papers. Both carry a medium repro-readiness rating: ArticleMiner's comparisons are candidly flagged because headline gains are confounded by extra evidence access, and the spectral smoothing paper reports modest gains with thin statistics. The latter is still judged plausible and drop-in, which supports third-party reimplementation.

Research summary

Jay Pujara is a researcher at the University of Southern California whose profile, name-resolved from OpenAlex bibliographic data, centers on knowledge graph technology. The snapshot splits evenly (0.5 / 0.5) between machine learning for knowledge graph representation learning and knowledge graph construction with scientific information extraction. Topic signals reinforce this: topic modeling (4), semantic web and ontologies (2), advanced graph neural networks (2), with single appearances in scientometrics and NLP techniques. Evolve has reviewed two 2026 arXiv papers. In ArticleMiner, an ontology-guided framework that converts scientific publications into knowledge graphs, Pujara is the last author; the review finds it beats a same-LLM few-shot baseline on four tasks while candidly noting the headline gains are confounded by extra evidence access. The second paper proposes spectral Laplacian initialization plus scheduled Dirichlet-energy smoothing for uncertain knowledge graph completion, improving over ssCDL; the verdict is plausible and drop-in, but gains modest and statistics thin. Output is recent and concentrated: 7 works in the last three years, with 1 counted in 2024 and 6 in 2026. Reproducibility averages 55 across the two reviewed papers, both rated medium. Collaboration is compact, with 4 distinct coauthors recorded and a score of 32 on that axis.

Collaboration pattern

This OpenAlex snapshot records only 4 distinct coauthors, and the collaboration dimension scores 32 accordingly — a small, dense network. Recurring partners are Craig A. Knoblock, Md Abrar Jahin, and Taufikur Rahman Fuad. Role placement on the two 2026 reviewed papers is senior-adjacent rather than lead: last author on ArticleMiner and a middle author on the spectral smoothing paper, with no corresponding-author credit on either.

Research keywords

knowledge graph constructionscientific information extractionuncertain knowledge graphsknowledge graph completionontology-guided extractionspectral initializationgraph Laplacian eigenvectorsDirichlet energy regularizationlarge language modelstable understandingtopic modelingsemantic web and ontologies

Highlights

ArticleMiner: ontology-guided paper-to-KG extraction reviewed at Evolve
In this 2026 arXiv paper Pujara is last author on an ontology-module framework that converts scientific publications, including tables and PDF content, into knowledge graphs. Evolve's review finds it beats a same-LLM few-shot baseline on four extraction tasks, while candidly noting the headline gains are confounded by extra evidence access; repro-readiness is medium.
Spectral initialization and scheduled smoothing for uncertain KG completion
The second 2026 reviewed paper, with Pujara in the middle of the author list, proposes spectral Laplacian-eigenvector initialization plus scheduled Dirichlet-energy smoothing for uncertain knowledge graph completion, improving over ssCDL. Evolve judges the method plausible and drop-in, but the gains modest and the statistics thin; repro-readiness is medium.
A two-track knowledge graph agenda, early-stage on record
The snapshot splits 0.5 / 0.5 between knowledge graph representation learning and KG construction with scientific information extraction, backed by topic counts in topic modeling (4) and advanced graph neural networks (2). Both reviewed 2026 papers rate medium for reproducibility with a mean of 55, marking a name-resolved, early-stage profile with output concentrated in 2026.

Timeline

2024Earliest year in this snapshot's countsByYear: 1 work recorded, marking the start of the visible output record.
2026Most active year on record: 6 works counted, including both Evolve-reviewed arXiv papers.
2026ArticleMiner appears on arXiv, an ontology-guided framework turning scientific publications into knowledge graphs, with Pujara as last author.
2026Spectral initialization and scheduled graph smoothing for uncertain knowledge graph completion released on arXiv; Pujara is a middle author.

Papers in the Evolve catalog

Frequent co-authors

Craig A. Knoblock×2 Md Abrar Jahin×1 Taufikur Rahman Fuad×1 Md Abrar Jahin×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 are drawn from an OpenAlex name-resolved identity; the recorded h-index of 0 and i10-index of 0 likely reflect incomplete identity resolution and should not be read as a true impact measure.
Only 2 papers carry Evolve reviews; radar scores rest on a thin base (reproPapers 2, distinctCoauthors 4).
countsByYear covers only 2024 and 2026; earlier career output is not represented in this snapshot.
Both reviewed papers are 2026 arXiv preprints; no peer-reviewed venue record is included in this scope.