Repository-wide refactor pilot
Branch: development
Status: Pilot reprocessing (review artifacts only — no corpus changes)
Date: 2026-06-14
Related: Repository-wide refactor governance · Repository-wide refactor audit · Source Packet Workflow
Purpose
Evaluate whether historically ingested grant-report and institutional program packets contain missed facilities, equipment, laboratories, centers, capabilities, relationships, and external partners that were overlooked under source-type-specific (grant-only) workflows.
This pilot does not modify the corpus (resources/, data/grants.yaml, people.yaml, publications, outputs, graph exports). Findings are indexed in:
data/review/resource_candidates.yamldata/review/relationship_candidates.yaml
Packets in scope
| Packet |
Primary source type (intake label) |
Prior workflow outcome |
2026-06-14-ors-annual-reports-2020-2025 |
ORS annual reports |
+826 grants, +133 PGL, +3 GRL |
2026-06-14-agif-reports-2018-2025 |
AGIF appendices |
+52 grants, +11 PGL, 0 GRL |
2026-06-14-rfums-dpu-grants-from-2024-and-2023 |
RFUMS collaborative awards |
+6 grants, PGL, EXT-011 prose |
2026-06-13-rfums-depaul-2025-grant-awardees-pdf |
RFUMS 2025 PDF |
+3 grants (2025 cohort) |
2026-06-13-dpu-rfu-awarded-projects |
DPU-RFU 2026 PDF |
+3 grants (merged) |
2026-06-14-circle-report-2024 |
Institutional program report |
Index only — no corpus merge |
Executive summary
| Metric |
Count |
| Total resource candidates |
52 |
| Total relationship candidates |
18 |
| Laboratories |
25 |
| Facilities / clinics |
6 |
| Centers / hubs |
6 |
| Equipment collections |
1 |
| Capabilities |
2 |
| Programs / initiatives |
6 |
| People (registry gaps) |
3 |
| Evidence-backed updates identified (not applied) |
1 summary + 6 high-confidence GRL gaps |
Finding: Grant-only intake captured award existence well but systematically dropped facility and lab signals, especially when registry pattern matching failed or no resource page existed. AGIF is the largest missed-facility cohort (0 GRL despite 18 lab/center/clinic titles). ORS applied 3 GRL from explicit pattern hits but left Connolly Lab, 52+ tier-C clinic/collaborative titles, and 141 tier-B rows without facility candidate indexing.
Classification key
| Class |
Meaning in this pilot |
| Evidence-backed update |
Sufficient evidence exists today for a corpus edit (not applied in this pilot) |
| Candidate resource |
Named lab/facility/center/program/equipment/capability insufficient for promotion |
| Candidate relationship |
Person↔facility, grant↔facility, or partner edge suggested but not evidenced enough |
| Rejected claim |
Source does not support the relationship (recorded in packet sections, not duplicated in YAML) |
Packet 1 — ORS annual reports (2026-06-14-ors-annual-reports-2020-2025)
Sources reviewed
- Six PDFs: FY 2020–2025 ORS Sponsored Research & Creative Activities annual reports (
uploads/)
- Extracted indices:
all_registry_awards_deduped.yaml, tier_a/b/c_*.yaml, facility_mention_hits.yaml, investigator_facility_edges.yaml, featured_extramural_awards.yaml
- Prior refactor output (grant-layer apply)
Full impact assessment
| Layer |
Reviewed |
Prior outcome |
Pilot finding |
| Grants |
Yes |
826 applied |
Complete for existence |
| People |
Partial |
133 PGL (resolved PI only) |
721 unresolved PI — expected |
| Facilities / labs |
Partial |
3 GRL |
Major gaps — see below |
| Equipment |
No |
— |
Not in report listings |
| Centers / programs |
No |
— |
NCTA, CGVRC, clinics missed |
| Publications |
No |
— |
Not primary in PDFs |
| External partners |
Partial |
CGVRC, MRSEC in titles |
Not indexed as EXT edges |
| Hyperlinks / reciprocity |
No |
— |
MedIX GRL not reciprocated in audit |
Missed facilities
| Name |
Evidence snippet |
Confidence |
Recommended action |
| Connolly Lab |
FY2020 featured NIH HSV: "The Connolly laboratory will provide significant expertise in the design, construction, and preliminary analysis of HSV-1 mutants" |
High |
Faculty/lab intake; create CSH resource or defer — do not invent page |
| DePaul Asylum and Immigration Legal Clinic |
Tier C listing title: College of Law Asylum and Immigration Law Clinic Legal… |
Medium |
Capture Law clinic official page; candidate facility |
| National Cybersecurity Teaching Academy (NCTA) |
Featured Sharevski NSF award narrative (GRA-050/051) |
Medium |
Program page intake; do not infer Secure Design Lab without naming |
Missed laboratories
| Name |
Evidence |
Confidence |
Match |
| Connolly Lab |
Featured narrative (Sarah Connolly PI) |
High |
No CSH page |
| Tier B lab/clinic/center titles |
141 registry-PI grants with lab words in title, facility_hits=[] |
Medium |
Various — re-scan needed |
| Tier C lab/clinic strings |
52+ unresolved listings (e.g. collaborative labs, clinics) |
Low–medium |
See resource_candidates.yaml RC-022+ |
Missed centers / initiatives
| Name |
Evidence |
Confidence |
| Chicago Gun Violence Research Collaborative (CGVRC) |
FY2020 listing: Sinai Urban Health Institute / CGVRC, $40,000 |
Medium |
| MRSEC: Center for Multifunctional Materials |
Tier C NSF collaborative title |
Low (likely external) |
| Chicago Alliance For Equity in Computer Science |
Multiple tier C collaborative titles |
Low |
Missed relationships
| Source → Target |
Evidence |
Confidence |
Action |
| Sarah Connolly → Connolly Lab |
Featured HSV text |
High |
Candidate until lab page exists |
| Filipo Sharevski → CDM-007 Secure Design Lab |
NCTA featured awards |
Low (rejected) |
Preserve as candidate — narrative does not name SDL |
| Daniela Raicu → CDM-011 MedIX |
Tier A REU |
High (applied) |
Verify CDM-011 reciprocal prose |
| Caitlin Karver → CSH-017 |
URC + listing |
Medium (applied) |
GRL-001 exists |
Evidence-backed updates (not applied)
- MedIX tier-A variants — GRL-013–015 already applied; audit whether additional MedIX-titled listings in tier B/C warrant GRL (classification:
evidence-backed_update summary in YAML)
Rejected claims
- Karver NIH featured (GRA-053) — correctly no facility edge (narrative omits lab name)
- Sharevski → Secure Design Lab — NCTA program name ≠ facility name
Packet 2 — AGIF reports (2026-06-14-agif-reports-2018-2025)
Sources reviewed
- Five AGIF annual report PDFs (2018–2025)
all_agif_projects_deduped.yaml (61 unique projects)- Prior apply: +52 grants, +11 PGL, 0 GRL
Full impact assessment
| Layer |
Prior outcome |
Pilot finding |
| Grants |
52 applied |
Complete |
| Facilities / labs |
0 GRL |
18 lab/center/clinic titles ignored |
| People |
11 PGL |
41 unresolved PI |
| Equipment |
— |
HPC Cloud → CDM-014 candidate |
| Programs |
— |
BETA Hub, School of Cinematic Arts marketing |
Missed facilities / labs (high value)
| Name |
PI |
Amount / FY |
Confidence |
Possible match |
| Immersion Lab |
David Ramsay (PER-001) |
$99,980 / 2024-25 |
Medium |
None — no registry page |
| CoBaAB Lab |
Thiru Ramaraj (PER-016) |
$100,000 / 2019-20 |
Medium |
None — distinct from CDM-011? |
| Virtual and Augmented Reality Communication Lab (VARC) |
Booth / McEwan |
$78,400 / 2019-20 |
Medium |
None |
| Cybersecurity Clinic |
Janine Spears (PER-008) |
$99,969 / 2023-24 |
Medium |
CDM-004 or CDM-007 (unconfirmed) |
| The Communication Clinic |
Maria de Moya Taveras |
$23,800 / 2020-21 |
Low |
None |
| Push Studio |
LeAnne Wagner (PER-023) |
$99,119 / 2020-21 |
Low |
None |
Missed facilities / labs (registry match — should have been GRL candidates)
| Name |
PI |
Confidence |
Match |
Recommended action |
| NexGeN Lab |
Sharief Oteafy (PER-004) |
High |
CDM-009 |
Add GRL after verify |
| Assisted Locomotion Laboratory |
Huzaifa (PER-007) |
High |
CDM-006 (RAL Lab) |
Add GRL — title matches existing page |
| DIGI Lab |
Jessica Klass |
High |
UNI-006 |
Add GRL — registry exists |
| High-Performance Compute Cloud |
Corin Picher |
High |
CDM-014 (HPCC) |
Verify scope; add GRL |
| DePaul Game Studio |
Allen Turner (fragment) |
Medium |
CDM-016 |
Capture AGIF line; verify title parse |
Missed centers / capabilities
| Name |
Evidence |
Confidence |
| Applied Theatrical Communication Center |
AGIF 2023-24, Dustin Goltz |
Low |
| BETA Hub |
AGIF 2023-24, James Mourey |
Low |
| Virtual Production |
AGIF 2020-21, Brian Andrews |
Low (capability / CDM production) |
Missed relationships
| Source → Target |
Confidence |
Action |
| AGIF NexGeN grant → CDM-009 |
High |
GRL candidate |
| AGIF Assisted Locomotion → CDM-006 |
High |
GRL candidate |
| AGIF DIGI → UNI-006 |
High |
GRL candidate |
| AGIF HPC Cloud → CDM-014 |
High |
GRL candidate |
| AGIF Immersion Lab → (new resource) |
Medium |
Lab intake first |
| AGIF CoBaAB → Ramaraj / new lab page |
Medium |
Disambiguate vs MedIX |
Rejected claims
- None at facility level — prior workflow silently omitted rather than rejected
Packet 3 — RFUMS 2023–2024 (2026-06-14-rfums-dpu-grants-from-2024-and-2023)
Sources reviewed
Summary_of_Awardees_22-23.pdf, Summary_of_DPU-RFU_Awarded_Projects_2024.pdf- Curator-added DePaul News captures (corroboration)
- Prior refactor: GRA-040–045, PGL, EXT-011 updates
Full impact assessment
| Layer |
Outcome |
| Grants / EXT-011 |
Evidence-backed — complete |
| People |
4 PGL; Pedram, Ustun, Hwang missing from registry |
| Facilities |
Rejected correctly — PDFs name colleges only |
Missed people (candidate resources)
| Name |
Role |
Evidence |
Confidence |
| Mahdi Pedram |
GRA-044 PI |
2024 SmokeSense award |
Medium |
| Ilyas Ustun |
GRA-041 co-PI |
2023 gait ML award |
Medium |
| Sungsoon (Julie) Hwang |
GRA-041 co-PI |
2023 award |
Medium |
Missed facilities / labs
None promoted — abstracts describe ML/gait/sepsis methods without naming DePaul labs. Prior refactor correctly rejected facility edges.
Candidate relationships (preserve, do not apply)
| Source → Target |
Evidence |
Confidence |
Action |
| PER-007 Huzaifa → CDM-006 RAL Lab |
RFUMS awards; CDM attribution only |
Low |
Do not infer from college name |
| PER-016 Ramaraj → CDM-011 MedIX |
RFUMS co-PI |
Low |
MedIX from other sources |
| PER-032 Furst → CDM-004 |
2024 sepsis award |
Low |
Rejected in prior refactor — preserve |
External partners
- RFUMS Resuscitation Institute mentioned in review notes — rejected (no DePaul facility evidence); preserved as external center candidate
Packet 4 — RFUMS 2025 (2026-06-13-rfums-depaul-2025-grant-awardees-pdf)
Sources reviewed
DPU-RFU_Awarded_Projects_2025.pdf- Prior intake: GRA-037–039, EXT-011, PGL for PER-052 only
Findings
| Category |
Result |
| Grants |
Evidence-backed — complete |
| Facilities / labs |
None in PDF |
| People |
Orhean, Ustun, Ansari, Alizadeh noted for profile follow-up |
| Relationships |
GRA-037–039 → EXT-011 — evidence-backed |
Pilot conclusion: Grant-only workflow appropriate for this packet; no facility omission detected.
Packet 5 — DPU-RFU 2026 (2026-06-13-dpu-rfu-awarded-projects)
Sources reviewed
Summary of DPU-RFU Awarded Projects 2026.pdfintake.yaml — merged status, EXT-011 target
Findings
| Category |
Result |
| Grants |
Three 2026 awards — evidence-backed (merged) |
| Facilities |
Explicitly excluded in do_not — PDF names colleges not labs |
| People |
PER-010, PER-016 identified; Jason/Wu unresolved |
Pilot conclusion: No facility signals missed; grant-layer scope was correct.
Packet 6 — CIRCLE report (2026-06-14-circle-report-2024)
Sources reviewed
CIRCLE_report_FINAL.pdfextracted/circle_summary.yaml- Prior intake: index only (
not_grant_intake: true)
Full impact assessment
| Layer |
Finding |
| Programs |
CIRCLE — Curriculum Integrating Research and Creative Learning in Education |
| Capabilities |
IRMA CIRCLE application (institutional analytics) |
| Courses |
Three course attribute types — crosswalk to CUR-* not performed |
| Facilities / equipment |
Not documented in report (correctly not inferred) |
| Grants / people |
N/A |
Missed programs / capabilities (candidate resources)
| Name |
Evidence snippet |
Confidence |
| CIRCLE program |
~50% of students enrolled in at least one research/creative inquiry course (2024) |
High |
| IRMA CIRCLE application |
Platform named in report for attribute tracking |
Medium |
| Course attributes (Creative Inquiry, Research, Research and Creative Inquiry) |
Taxonomy in report tables |
Medium |
Missed relationships
| Source → Target |
Evidence |
Confidence |
| CIRCLE attributes → CUR-* courses |
Participation measured via attributed courses |
Low |
| CIRCLE → ORS research culture |
Complementary institutional metrics |
Low (narrative only) |
Rejected claims
- Facility or equipment ownership from CIRCLE report — report does not document hardware (prior
intake_circle_report.py scope limit was correct for facilities, but program layer was under-indexed)
Aggregate summary
| Category |
Count |
| Total resource candidates |
52 |
| Total relationship candidates |
18 |
| Laboratories |
25 |
| Facilities (incl. clinics) |
6 |
| Centers / hubs |
6 |
| Equipment collections |
1 |
| Capabilities |
2 |
| Programs / initiatives / research groups |
8 |
| People registry gaps |
3 |
By packet (resource candidates)
| Packet |
Candidates |
| ORS annual reports |
~28 |
| AGIF reports |
~14 |
| RFUMS 2023–2024 |
~4 |
| RFUMS 2025 |
0 |
| DPU-RFU 2026 |
0 |
| CIRCLE |
~6 |
Highest-priority follow-up (implementation order)
- AGIF → registry GRL gaps — NexGeN (CDM-009), RAL Lab (CDM-006), DIGI (UNI-006), HPCC (CDM-014) — high confidence, existing pages
- Connolly Lab — ORS featured narrative + no CSH page — faculty/lab intake packet
- Immersion Lab / CoBaAB / VARC — AGIF-named labs without registry pages — lab website capture
- ORS tier B re-scan — 141 grants with lab/clinic/center in title, zero
facility_hits
- CIRCLE program indexing — institutional program record + course-attribute crosswalk (separate curriculum packet)
- RFUMS recurring PIs — Pedram, Ustun, Hwang profile packets (people layer, not facilities)
Conclusion
Earlier grant-only workflows did not lose award knowledge (ORS/AGIF grant counts are substantial) but did lose cross-layer institutional knowledge:
- Facilities and labs named in AGIF project titles were not preserved as candidates or GRL edges.
- ORS applied narrow regex facility patterns (14 patterns) and indexed only 4 facility mention hits plus 3 GRL while 721 unresolved and 141 tier-B rows received no facility candidate file.
- RFUMS/DPU-RFU packets appropriately stayed grant-focused — minimal missed facility signal.
- CIRCLE was under-indexed on program/course layers, not facilities.
A broader repository-wide reprocessing effort is justified for ORS and AGIF packets; RFUMS/DPU-RFU packets need only people follow-up; CIRCLE needs program-layer indexing.
Artifacts
| File |
Contents |
data/review/resource_candidates.yaml |
52 candidate resources with provenance |
data/review/relationship_candidates.yaml |
18 candidate relationships |
No corpus files modified. Stop before commit.