README v0 10 6
# HE Open Data Ranking 2027 — v0.10.6 Candidate-Final Package
## Purpose
v0.10.6 is the frozen technical basis of HE Open Data Academic Ranking 2027 Methodology v1.0. This full package preserves the v0.10.6 collector and scoring architecture and adds the optional Researcher-Level Institutional Intelligence v1.0 dashboard layer without changing the official university score.
## Current scoring architecture
- 10,000 total points
- 5 pillars
- 24 KPIs
- Research Performance & Open Knowledge: 50% / 5,000 points
- Internationalization & Global Networks: 18% / 1,800 points
- Research Capacity & Academic Breadth: 15% / 1,500 points
- Societal Relevance & Sustainable Research: 7% / 700 points
- Cross-Sector Research Engagement: 10% / 1,000 points
- Open Knowledge Infrastructure (KPI8): 600 points, unchanged
## v0.10.6 methodology refinements
The collector keeps the raw evidence architecture and Eligibility unchanged. The scorer includes the calibrated weights above and retains the three de-correlation definitions introduced during validation:
1. KPI10 — Multilateral Research Collaboration
- measures 3+ country works as a share of international works rather than total works;
- reliability-adjusted; zero when there are no international works.
2. KPI21 — High-Impact SDG Research
- measures SDG Top-10% excellence advantage relative to the institution's own overall Top-10% research baseline.
3. KPI22 — Open SDG Research
- measures SDG open-access advantage relative to the institution's own overall OA baseline.
No institutional entity-boundary aggregation has been added. Eligibility remains exactly under HE control through the existing override mechanism.
## OpenAlex API key decision
v0.10.6 intentionally uses a single `OPENALEX_API_KEY`.
A second key is **not** used to speed collection because the collector's OpenAlex backbone uses sequential aggregate/count/group queries. Previous timing data showed only about 2–3 OpenAlex requests per second on average, far below OpenAlex's documented request-rate ceiling. A second key therefore would not reduce request latency or make the sequential steps execute in parallel. It could only change available account budget/quota, which is a different issue from speed.
The collector already runs independent secondary sources in parallel where safe (Crossref, DataCite, OpenAIRE, OpenAlex Sources after the primary OpenAlex stage).
## Input
The collector reads `RORDOMAIN.csv` from the package folder.
Required columns:
- `id`
- `links.type.website`
Do not rename your real input to `RORDOMAIN_TEMPLATE.csv`; the template is only an example.
## First-time setup
1. Install Python 3.10+.
2. Run: `pip install -r requirements_v0_10_6.txt`
3. Copy `.env.example` to `.env` and add your credentials.
4. Place your actual `RORDOMAIN.csv` in this folder.
## Recommended run sequence
1. `01_PREPARE_ACADEMIC_CACHES_WINDOWS_v0_10_6.bat`
2. `02_RUN_COLLECTION_RESUME_WINDOWS_v0_10_6.bat`
3. Optional status check: `03_SHOW_COLLECTION_STATUS_WINDOWS_v0_10_6.bat`
4. Optional raw export: `04_EXPORT_CURRENT_METRICS_WINDOWS_v0_10_6.bat`
5. After collection is complete: `05_RUN_SCORING_WINDOWS_v0_10_6.bat`
## Default collection outputs
- `HE_Open_Data_Ranking_2027_Collection_v0_10_6.sqlite`
- `HE_Open_Data_Ranking_2027_Metrics_Master_v0_10_6.xlsx`
- `progress_v0_10_6.json`
- `progress_checklist_v0_10_6.csv`
- `academic_cache_v0_10_6/`
The raw master workbook includes `Groupings` / `Groupings_###` sheets, which are required for native KPI20 SDG Diversity calculation. A fresh v0.10.6 run therefore does not need a prior ranking workbook as a reference.
## Default scoring output
- `HE_Open_Data_Ranking_2027_Ranking_v0_10_6.xlsx`
## Researcher-Level Institutional Intelligence v1.0
This package now includes an optional **non-scoring** researcher-intelligence layer:
- `he_researcher_intelligence_v1_0.py`
- HE Institutional Research Contribution Score (HE-IRCS)
- up to 100 **Leading Research Contributors** per university
- exact author+institution attribution at the OpenAlex authorship level
- API pilot mode and local OpenAlex JSONL snapshot production mode
- SQLite, XLSX, CSV and JSON outputs for dashboard integration
HE-IRCS uses institution-attributable metrics only for ordering the researcher list:
- 40% affiliated citation impact percentile
- 25% affiliated h-index percentile
- 15% affiliated i10-index percentile
- 10% affiliated publications percentile
- 10% recent institutional citation impact percentile (2021–2025)
This module **does not alter** the official HE university score, any of the 24 KPIs, or the 10,000-point architecture. It is intentionally separated from the frozen Methodology v1.0.
See `README_RESEARCHER_INTELLIGENCE_v1_0.md` and `RESEARCHER_INTELLIGENCE_METHODOLOGY_v1_0.md`.
### Researcher intelligence run sequence
After the core collector has produced `HE_Open_Data_Ranking_2027_Collection_v0_10_6.sqlite`:
- Pilot / small batch: `06_RUN_RESEARCHER_INTELLIGENCE_API_PILOT_WINDOWS_v1_0.bat`
- Production / large cohort: `06_RUN_RESEARCHER_INTELLIGENCE_SNAPSHOT_WINDOWS_v1_0.bat`
- Status: `07_SHOW_RESEARCHER_INTELLIGENCE_STATUS_WINDOWS_v1_0.bat`
- Export-only: `08_EXPORT_RESEARCHER_INTELLIGENCE_WINDOWS_v1_0.bat`
- Recalculate HE-IRCS after a full collection: `09_RECALCULATE_RESEARCHER_SCORES_WINDOWS_v1_0.bat`
For 9,500+ universities, the local OpenAlex snapshot route is strongly preferred to the live API.
## Resume safety
Every completed institution is committed transactionally to SQLite. If the process is interrupted, rerun the same collection batch file. Completed institutions are skipped automatically and the first unfinished institution is retried.
## Eligibility
No eligibility rule was changed. Every input institution remains included unless HE explicitly sets `Decision=EXCLUDE` in `eligibility_overrides.csv`.
## Validation / freeze status
The v0.10.6 core scoring architecture is the frozen technical basis for Methodology v1.0. Do not retune 2027 indicator definitions or weights merely to improve convergence with external rankings or to change individual institutional positions. The Researcher Intelligence v1.0 module is non-scoring and therefore does not alter the methodology freeze.
## Purpose
v0.10.6 is the frozen technical basis of HE Open Data Academic Ranking 2027 Methodology v1.0. This full package preserves the v0.10.6 collector and scoring architecture and adds the optional Researcher-Level Institutional Intelligence v1.0 dashboard layer without changing the official university score.
## Current scoring architecture
- 10,000 total points
- 5 pillars
- 24 KPIs
- Research Performance & Open Knowledge: 50% / 5,000 points
- Internationalization & Global Networks: 18% / 1,800 points
- Research Capacity & Academic Breadth: 15% / 1,500 points
- Societal Relevance & Sustainable Research: 7% / 700 points
- Cross-Sector Research Engagement: 10% / 1,000 points
- Open Knowledge Infrastructure (KPI8): 600 points, unchanged
## v0.10.6 methodology refinements
The collector keeps the raw evidence architecture and Eligibility unchanged. The scorer includes the calibrated weights above and retains the three de-correlation definitions introduced during validation:
1. KPI10 — Multilateral Research Collaboration
- measures 3+ country works as a share of international works rather than total works;
- reliability-adjusted; zero when there are no international works.
2. KPI21 — High-Impact SDG Research
- measures SDG Top-10% excellence advantage relative to the institution's own overall Top-10% research baseline.
3. KPI22 — Open SDG Research
- measures SDG open-access advantage relative to the institution's own overall OA baseline.
No institutional entity-boundary aggregation has been added. Eligibility remains exactly under HE control through the existing override mechanism.
## OpenAlex API key decision
v0.10.6 intentionally uses a single `OPENALEX_API_KEY`.
A second key is **not** used to speed collection because the collector's OpenAlex backbone uses sequential aggregate/count/group queries. Previous timing data showed only about 2–3 OpenAlex requests per second on average, far below OpenAlex's documented request-rate ceiling. A second key therefore would not reduce request latency or make the sequential steps execute in parallel. It could only change available account budget/quota, which is a different issue from speed.
The collector already runs independent secondary sources in parallel where safe (Crossref, DataCite, OpenAIRE, OpenAlex Sources after the primary OpenAlex stage).
## Input
The collector reads `RORDOMAIN.csv` from the package folder.
Required columns:
- `id`
- `links.type.website`
Do not rename your real input to `RORDOMAIN_TEMPLATE.csv`; the template is only an example.
## First-time setup
1. Install Python 3.10+.
2. Run: `pip install -r requirements_v0_10_6.txt`
3. Copy `.env.example` to `.env` and add your credentials.
4. Place your actual `RORDOMAIN.csv` in this folder.
## Recommended run sequence
1. `01_PREPARE_ACADEMIC_CACHES_WINDOWS_v0_10_6.bat`
2. `02_RUN_COLLECTION_RESUME_WINDOWS_v0_10_6.bat`
3. Optional status check: `03_SHOW_COLLECTION_STATUS_WINDOWS_v0_10_6.bat`
4. Optional raw export: `04_EXPORT_CURRENT_METRICS_WINDOWS_v0_10_6.bat`
5. After collection is complete: `05_RUN_SCORING_WINDOWS_v0_10_6.bat`
## Default collection outputs
- `HE_Open_Data_Ranking_2027_Collection_v0_10_6.sqlite`
- `HE_Open_Data_Ranking_2027_Metrics_Master_v0_10_6.xlsx`
- `progress_v0_10_6.json`
- `progress_checklist_v0_10_6.csv`
- `academic_cache_v0_10_6/`
The raw master workbook includes `Groupings` / `Groupings_###` sheets, which are required for native KPI20 SDG Diversity calculation. A fresh v0.10.6 run therefore does not need a prior ranking workbook as a reference.
## Default scoring output
- `HE_Open_Data_Ranking_2027_Ranking_v0_10_6.xlsx`
## Researcher-Level Institutional Intelligence v1.0
This package now includes an optional **non-scoring** researcher-intelligence layer:
- `he_researcher_intelligence_v1_0.py`
- HE Institutional Research Contribution Score (HE-IRCS)
- up to 100 **Leading Research Contributors** per university
- exact author+institution attribution at the OpenAlex authorship level
- API pilot mode and local OpenAlex JSONL snapshot production mode
- SQLite, XLSX, CSV and JSON outputs for dashboard integration
HE-IRCS uses institution-attributable metrics only for ordering the researcher list:
- 40% affiliated citation impact percentile
- 25% affiliated h-index percentile
- 15% affiliated i10-index percentile
- 10% affiliated publications percentile
- 10% recent institutional citation impact percentile (2021–2025)
This module **does not alter** the official HE university score, any of the 24 KPIs, or the 10,000-point architecture. It is intentionally separated from the frozen Methodology v1.0.
See `README_RESEARCHER_INTELLIGENCE_v1_0.md` and `RESEARCHER_INTELLIGENCE_METHODOLOGY_v1_0.md`.
### Researcher intelligence run sequence
After the core collector has produced `HE_Open_Data_Ranking_2027_Collection_v0_10_6.sqlite`:
- Pilot / small batch: `06_RUN_RESEARCHER_INTELLIGENCE_API_PILOT_WINDOWS_v1_0.bat`
- Production / large cohort: `06_RUN_RESEARCHER_INTELLIGENCE_SNAPSHOT_WINDOWS_v1_0.bat`
- Status: `07_SHOW_RESEARCHER_INTELLIGENCE_STATUS_WINDOWS_v1_0.bat`
- Export-only: `08_EXPORT_RESEARCHER_INTELLIGENCE_WINDOWS_v1_0.bat`
- Recalculate HE-IRCS after a full collection: `09_RECALCULATE_RESEARCHER_SCORES_WINDOWS_v1_0.bat`
For 9,500+ universities, the local OpenAlex snapshot route is strongly preferred to the live API.
## Resume safety
Every completed institution is committed transactionally to SQLite. If the process is interrupted, rerun the same collection batch file. Completed institutions are skipped automatically and the first unfinished institution is retried.
## Eligibility
No eligibility rule was changed. Every input institution remains included unless HE explicitly sets `Decision=EXCLUDE` in `eligibility_overrides.csv`.
## Validation / freeze status
The v0.10.6 core scoring architecture is the frozen technical basis for Methodology v1.0. Do not retune 2027 indicator definitions or weights merely to improve convergence with external rankings or to change individual institutional positions. The Researcher Intelligence v1.0 module is non-scoring and therefore does not alter the methodology freeze.