Publications quality review
Review date: 2026-05-31
Scope: Cohort 4 publication discovery batch (2026-05-31) + full-layer audit of publication–resource links
Sources: full-refresh-report.md, publication-intake-rules.md, data/publications.yaml, link YAML, data/graph_exports/*.csv
Cohort 4 batch detail is summarized in §1 below; historical pilot/cohort execution logs live in _local_archive/reports/ (local only, not committed).
1. Batch summary (cohort 4)
| Metric | Count |
|---|---|
| People searched | 10 |
| Candidate publications found (approx.) | 12 |
| Included in registry | 9 |
| Maybe (deferred) | 3 |
| Excluded | 1 |
| New person–publication links | 9 (PPL-031 … PPL-039) |
| New publication–resource links | 0 |
Deferred (maybe): Adelakun AMCIS 2023 iD Lab paper; Bennett Frontiers editorial; Schrank BragFish (pre-DePaul).
Excluded: PER-112 Christine Binder — no peer-reviewed scholarly output found.
Post-batch registry: 34 publications · 39 person–publication links · 6 publication–resource links (unchanged count).
2. Cohort 4 publication relevance classification
| pub_id | person | Title (short) | Relevance | Rationale |
|---|---|---|---|---|
| PUB-026 | Furst | DiiS biomedical FAIR framework | Strong | DePaul coauthor; medical informatics / distributed data; aligns with GRA-016 federated-learning thread and VIDA adjacency. Person–pub only. |
| PUB-027 | Putnam | UX accessibility (ACM TAACCESS) | Strong | DePaul HCI faculty; matches GRA-013 UX pedagogy grant; hard-tech HCI methods research. |
| PUB-028 | Wagner | COVID vaccination qualitative study | Weak | Authorship verified, but Clinical Nursing Research focus-group study does not document SPARK data-viz, instrumentation, or hard-tech. PPL-033 set include_in_graph: false. |
| PUB-029 | Adelakun | Telemedicine satisfaction framework | Strong | DePaul coauthor; health informatics / IS quality; iD Lab director research thread. |
| PUB-030 | Oteafy | Remote surgery / tactile internet (ICC) | Strong | DePaul-era IoT/CPS standards work; NexGeN Lab faculty lead; explicit hard-tech networking topic. |
| PUB-031 | Bennett | Digital health EMA / predictive modeling | Medium | DePaul author; ML/digital-health methods in title. Topical bridge to CyberLabs/MedIX in relevance text only — no pub–resource edge (correct). |
| PUB-032 | Schrank | Avant-garde Videogames (book) | Medium | Valid DePaul scholarly output for VAD Lab director; game-theory/cultural analysis, not instrumentation or lab hardware documentation. Acceptable person–pub at pilot bar. |
| PUB-033 | Bystriansky | Wolf eel salinity physiology | Medium | Strong bench-methods paper; species differs from GRA-017 Poeciliid grant but same osmoregulation theme. No McGowan facility text. |
| PUB-034 | Kozlowski | Repeat subconcussion (rats) | Strong | Direct match to GRA-010 manuscript grant; rodent neuro methods on faculty profile. |
Summary counts (cohort 4 only)
| Class | Count |
|---|---|
| Strong | 5 |
| Medium | 3 |
| Weak | 1 |
| Questionable | 0 |
3. Accepted strong records (graph-included)
Person–publication links retained with include_in_graph: true and high confidence:
- PUB-026 / PPL-031 (Furst)
- PUB-027 / PPL-032 (Putnam)
- PUB-029 / PPL-034 (Adelakun)
- PUB-030 / PPL-035 (Oteafy)
- PUB-034 / PPL-039 (Kozlowski)
Publication–resource links with explicit facility evidence (pre-cohort, still graph-included):
- PRL-PUB-001 — PUB-002 → EXT-003 (APS trial run in abstract)
- PRL-PUB-002 — PUB-010 → EXT-003 (DePaul news: APS data collection)
- PRL-PUB-005 — PUB-013 → CSH-017 (Karver profile: assay developed in lab)
- PRL-PUB-006 — PUB-013 → CSH-018 (Southern profile: joint lab development)
4. Records retained but weak / medium (graph-included)
| Record | Disposition | Notes |
|---|---|---|
| PUB-031 / PPL-036 (Bennett) | Keep in graph | Digital-health ML content; monitor for stronger MedIX/CyberLabs facility paper |
| PUB-032 / PPL-037 (Schrank) | Keep in graph | Scholarly book, not instrumentation; documents faculty research program not lab hardware |
| PUB-033 / PPL-038 (Bystriansky) | Keep in graph | Bench physiology; grant species mismatch is topical only |
5. Records changed to include_in_graph: false (this review)
| Link | Change | Reason |
|---|---|---|
| PPL-033 Wagner → PUB-028 | include_in_graph: false, confidence → medium |
Qualitative nursing study; no hard-tech / SPARK / instrumentation signal |
| PRL-PUB-003 PUB-006 → CSH-009 | include_in_graph: false |
Balloon proceedings linked to STEM Center via profile/contact page only — not paper text |
| PRL-PUB-004 PUB-009 → CSH-016 | include_in_graph: false, confidence → medium |
Profile method-list alignment; no explicit Grice Lab / McGowan acknowledgment in publication |
Publication PUB-028 confidence downgraded to medium in publications.yaml; relevance text corrected to avoid SPARK infrastructure overclaim.
6. Records flagged needs_review (no removal)
| Item | Issue | Recommended action |
|---|---|---|
| PUB-028 | Retained in registry for authorship audit trail | OK to keep YAML record; graph edge excluded |
| PRL-PUB-003, PRL-PUB-004 | Pre-cohort weak pub–resource edges | Revisit if full paper PDF / methods skim finds facility acknowledgment |
| PUB-020 (Kipp, pre-cohort) | Clinical imaging without lab methods text | Acceptable person–pub; no pub–resource edge (correct) |
| PUB-005 (Rosing) | Community food-systems, not hardware | Acceptable at pilot bar; no pub–resource edge |
No records removed from registry in this pass.
7. Publication–resource link audit (all 6)
| link_id | pub → resource | Explicit evidence? | Verdict |
|---|---|---|---|
| PRL-PUB-001 | PUB-002 → EXT-003 | Yes — APS trial run in abstract | Valid |
| PRL-PUB-002 | PUB-010 → EXT-003 | Yes — DePaul news quotes APS data collection | Valid |
| PRL-PUB-003 | PUB-006 → CSH-009 | No — profile + center contact only | Downgraded (topic/affiliation) |
| PRL-PUB-004 | PUB-009 → CSH-016 | No — profile method alignment only | Downgraded (weak lab association) |
| PRL-PUB-005 | PUB-013 → CSH-017 | Yes — profile states assay developed in Karver lab for this work | Valid (profile names specific paper/lab) |
| PRL-PUB-006 | PUB-013 → CSH-018 | Yes — Southern profile joint FRET development with Karver | Valid |
Cohort 4 added zero publication–resource links (correct policy).
8. Author-identity ambiguity
| Person | Risk | Resolution |
|---|---|---|
| Casey Bennett (PER-011) | Common name in clinical ML literature | DePaul affiliation on TRJ 2026 author list; faculty profile match — low risk |
| Jacob Furst (PER-052) | Moderate surname collision | Crossref uses Jacob D. Furst, DePaul CDM — low risk |
| LeAnne Wagner (PER-023) | Spelling variants | PMC author list + DePaul affiliation — low risk |
| Olayele Adelakun (PER-017) | Yele / Olayele variants on Scholar | Crossref DePaul affiliation on JMS 2019 — low risk |
| Dorothy Kozlowski (PER-045) | Uncommon in neuroscience | Faculty profile + grant GRA-010 manuscript pointer — low risk |
No homonym collisions requiring confidence: low on included graph edges.
9. Cross-unit collaboration implications
Person–publication edges do not create person–person collaboration edges. Review found no new misleading cross-unit implications from cohort 4:
- Wagner coauthors (nursing faculty) do not imply CDM↔nursing lab collaboration in the resource graph.
- Furst DiiS coauthor network (Raicu) is not exported as person–person edges.
- Oteafy ICC external coauthors do not create DePaul↔external facility edges.
No false hub inflation from publications in cohort 4 — zero new pub–resource links. Hub resources (CDM-015, CSH-017, etc.) remain driven by person–resource and prior pub–resource edges only.
10. False hub check
| Resource | Pub–resource degree | Assessment |
|---|---|---|
| CSH-017 | 1 (PUB-013) | Justified by explicit Karver lab profile evidence |
| CSH-018 | 1 (PUB-013) | Justified by explicit Southern/Karver collaboration evidence |
| EXT-003 | 2 (PUB-002, PUB-010) | Both have explicit APS use text |
| CSH-009 | 0 after downgrade | No longer inflated by weak balloon link |
| CSH-016 | 0 after downgrade | No longer inflated by profile-inference link |
Publications layer does not currently create misleading resource hubs after downgrades.
11. Recommendations before next batch
- Tighten intake filter for center directors — require instrumentation, computing, sensing, fabrication, or explicit methods/lab text; exclude pure qualitative health/social-science coauthorship unless paper documents mapped resources.
- SPARK / design directors — prioritize papers mentioning data visualization, predictive modeling pipelines, or civic analytics methods; not general HCD coauthorship on unrelated domain studies.
- Theatre design-tech (TTS) — Binder pass found no scholarly layer; next TTS candidates (Kelly, Farina) need USITT/proceedings search with same conservative bar, or defer TTS until non-peer outputs are scoped separately.
- Publication–resource links — continue PDF/methods skim before any new pub–resource edge; never infer from faculty profile method lists alone (PRL-PUB-004 lesson).
- Grant alignment — use grant abstract as search pointer, not as sufficient evidence for resource linkage (Bystriansky wolf eel vs Poeciliid grant is acceptable person–pub, not pub–resource).
- Batch size — 8–10 people remains appropriate; expect ~1 weak inclusion per batch at current coverage — plan a quality review pass after each intake.
- Re-audit pre-cohort pub–resource edges if full-text methods review becomes available for PUB-006 and PUB-009.
12. Graph state after corrections
Edits applied in this review:
PPL-033:include_in_graph: falsePRL-PUB-003,PRL-PUB-004:include_in_graph: falsePUB-028: confidence → medium, relevance text corrected
Regenerate: python3 scripts/refresh_all.py
Historical pilot and cohort batch logs: _local_archive/reports/ (local only).