Documentation — G8 Research Output Database

1. Data Sources & Joins

Each record combines three kinds of source: the university's official staff list, the university's own publication records supplemented by open indexes, and journal ranking lists.

Researcher profiles

For every university, the staff list comes from the official Accounting and Finance staff directory. Each researcher's profile provides their name, job title (from which the academic level is derived), field of research, profile URL and, where the university publishes them, an ORCID iD and a repository ID (source_id). Staff with no publications are left out.

Publications

Publications come first from each university's own system, which is treated as the official record:

UQ — UQ eSpace, looked up by each person's eSpace author ID.
UWA — UWA's Pure research portal: each person's publication feed and publication pages.
Monash — Monash's Pure research portal: each person's publication feed. OpenAlex then adds DOIs and ISSNs to the portal's papers by exact title match, using only OpenAlex records whose ORCID matches the researcher.
Melbourne — Minerva Access, the University of Melbourne repository.
ANU, UNSW and Sydney — each researcher's official staff profile page.
Adelaide — the journal articles listed on each researcher's Adelaide research profile, each with its DOI.

Three open indexes then add papers the university's records miss: ORCID (the researcher's own public record), Crossref and OpenAlex. They are looked up by ORCID iD, and the OpenAlex search is restricted to papers written while at that university. Duplicate copies of the same paper are merged by DOI, or by title and year where there is no DOI. Only journal articles are kept.

Co-authored papers are counted once per matching author, so a paper with two staff co-authors appears twice in the database.

Journal rankings

Journal quality data comes from three sources, matched to publications by ISSN (the ABDC list falls back to an exact title match when a paper has no ISSN):

ABDC Journal Quality List 2025 — from data/ABDC-JQL-2025-v1-260326.xlsx. Provides quality_rank (A*, A, B, C) and abdc_edition.
Scimago Journal Rankings 2025 — from data/scimagojr 2025.csv. Provides sjr, sjr_quartile, h_index and cites_per_doc_2y.
Clarivate Journal Impact Factor (JCR) — fetched from the Clarivate Journal Citation Reports API when an API key is configured. Provides impact_factor, impact_factor_5yr and jcr_year. Only journals indexed in Web of Science have an impact factor; for the rest these are null.

2. Standardisation Pipeline

One command runs the whole pipeline for a university: python run.py --uni <name>, or python run.py --all for all eight. Each university goes through the same steps:

Step 1 — Collect

The university's adapter (base_scrapers/<name>.py) reads the official staff directory and the university's own publication records. Reviewed identity corrections in data/ (for example a verified ORCID) are applied here.

Step 2 — Supplement

ORCID, Crossref and OpenAlex are searched for papers the university's records miss.

Step 3 — Clean and enrich

Titles, years, DOIs and publication types are cleaned; errata, editorials and reviewed exclusions (data/publication_exclusions.csv) are removed. OpenAlex and Crossref then add citation metrics, open-access status and missing ISSNs by DOI.

Step 4 — Rate journals

Each journal is matched to the ABDC list, Clarivate Journal Citation Reports (when an API key is configured) and Scimago.

Step 5 — Screen

screen.py removes open-index papers that probably belong to someone else with the same name: when a researcher has at least five open-index journal articles and fewer than a quarter of them are in ABDC-rated journals, those papers are set aside. Some specific cases are also handled by hand-reviewed rules. Papers from the university's own records are never removed. Everything set aside is listed in <uni>_screened_out.csv with the reason.

Step 6 — Export

The results are written to final output/<uni>/ as four tables — staff, publications, journals and harvest — each as CSV and JSON.

Step 7 — Load

load.py rebuilds the website's SQLite database (site/research.db) from every university's folder in final output/. The admin page's Refresh button runs the pipeline for one or all universities and then this step, replacing the live database only if everything succeeds.

OpenAlex percentile

Where available, OpenAlex returns a citation_normalized_percentile value indicating a publication's citation standing among comparable works. This is stored as citation_percentile in the database.

3. Database Schema

The database (site/research.db) is a SQLite file managed by SQLAlchemy and rebuilt by load.py. It has three tables in use, below. A fourth, harvest, is defined but not currently filled; when each source was last harvested is recorded in each university's <uni>_harvest file instead.

Researcher

Column	Type	Notes
`researcher_id`	INTEGER PK	Auto-increment
`name`	TEXT	Name with titles removed
`job_title`	TEXT	Position as the university lists it; the staff CSV adds `academic_title` (from the level) and `admin_title` (e.g. "Dean")
`academic_level`	TEXT	A–E, derived from the job title; null when it doesn't map
`university`	TEXT	University name (see §8)
`field_of_research`	TEXT	Accounting or Finance
`source_id`	TEXT	ID in the university's own system, where published
`orcid`	TEXT	ORCID iD, where available
`profile_url`	TEXT	Link to the official university profile

Journal

Column	Type	Notes
`journal_id`	INTEGER PK
`journal_name`	TEXT	ABDC title where listed, otherwise the name as found; unique
`journal_raw`	TEXT	Name exactly as the source gave it
`publisher`	TEXT
`issn`	TEXT	One or more ISSNs, separated by `;`
`quality_rank`	TEXT	ABDC: A*, A, B or C; null if not on the list (see §4)
`abdc_edition`	TEXT	E.g. "2025"
`impact_factor`	REAL	Clarivate 2-year JIF; null outside Web of Science
`impact_factor_5yr`	REAL	Clarivate 5-year JIF
`jcr_year`	INTEGER	JCR release the impact factors come from
`sjr`	REAL	Scimago Journal Rank score
`sjr_quartile`	TEXT	Q1–Q4 per Scimago
`h_index`	INTEGER	Journal h-index per Scimago
`cites_per_doc_2y`	REAL	Scimago 2-year citations per document
`scimago_year`	TEXT	Scimago release the figures come from

Publication

Column	Type	Notes
`publication_id`	INTEGER PK
`researcher_id`	INTEGER FK	→ researcher.researcher_id
`journal_id`	INTEGER FK	→ journal.journal_id (nullable)
`title`	TEXT
`year`	INTEGER
`author_count`	INTEGER	Including external co-authors
`authors`	TEXT	All authors, separated by `;`
`doi`	TEXT
`article_url`	TEXT	DOI link, or the source's link when there is no DOI
`link`	TEXT	Link given by the source
`quality_rank`	TEXT	The journal's ABDC rating, copied for convenience
`sjr_quartile`	TEXT	The journal's Scimago quartile, copied for convenience
`citation_percentile`	REAL	OpenAlex citation percentile, 0–1
`cited_by_count`	INTEGER	From OpenAlex
`fwci`	REAL	OpenAlex Field-Weighted Citation Impact
`oa_status`	TEXT	gold, diamond, hybrid, bronze, green or closed
`oa_url`	TEXT	Link to a free copy, where one exists
`publication_status`	TEXT	published or forthcoming (UWA rows use UWA's own wording, e.g. "Published - Jun 2024")
`source`	TEXT	The university system or open index the record came from

4. Ranking Metrics

ABDC Journal Quality List

The Australian Business Deans Council (ABDC) Journal Quality List rates journals in business and economics. The 2025 edition is used. The quality_rank field takes one of four values, or is blank when the journal is not on the list — common and expected for journals outside business, such as medical or engineering journals:

A*Top 7% of journals in the field. Highest quality.

AHigh quality, well regarded by the field.

BRecognised quality journals in the field.

CRecognised journals, lower tier.

Scimago Journal Rankings (SJR)

Scimago provides three metrics sourced from Elsevier Scopus data:

sjr — weighted citation score; accounts for the prestige of the citing journals.
sjr_quartile — Q1 (top 25%) to Q4 (bottom 25%) within the journal's subject category.
cites_per_doc_2y — average citations per document over the preceding two years (analogous to a 2-year impact factor).

Clarivate Journal Impact Factor (JCR)

impact_factor is the Clarivate 2-year Journal Impact Factor from Web of Science Journal Citation Reports, and impact_factor_5yr the 5-year version; jcr_year is the JCR release they come from. They are null for journals outside Web of Science. Scimago's cites_per_doc_2y is a similar measure with wider coverage.

OpenAlex Citation Percentile

citation_percentile is OpenAlex's citation percentile for the individual publication, not the journal, normalised by publication year, type and subfield. It is stored as a fraction from 0 to 1: a value of 0.95 means the paper is cited more than 95% of comparable papers.

5. Column Guide

A short guide to the columns in the data. A blank value always means not available, never zero: a journal with no ABDC rating is simply not on the ABDC list.

Researchers

Column	Meaning
`academic_level`	Australian academic level from the job title: A Associate Lecturer, B Lecturer, C Senior Lecturer, D Associate Professor, E Professor.
`field_of_research`	Accounting or Finance, from the department the university lists the person under.
`orcid`	A permanent researcher identifier, used to find their publications reliably.

Publications

Column	Meaning
`cited_by_count`	How many times the article has been cited.
`fwci`	Field-Weighted Citation Impact: citations compared with similar articles of the same field, year and type. 1.0 is the world average; 2.0 is twice as many citations as expected.
`citation_percentile`	Where the article ranks by citations among comparable articles, from 0 to 1. 0.91 means cited more than about 91% of them.
`oa_status`	Whether it is free to read: gold, diamond, hybrid or bronze (free from the publisher), green (a free copy in a repository), or closed.
`source`	Where the record came from: the university's own system (the official record) or an open index (ORCID, Crossref, OpenAlex).

Journals

Column	Meaning
`quality_rank`	ABDC rating: A* (highest), A, B or C.
`impact_factor`	Clarivate Journal Impact Factor: average citations in a year to the journal's articles from the previous two years. Only journals in Web of Science have one.
`impact_factor_5yr`	The same over five years; steadier, and better suited to slower-citing fields such as accounting.
`sjr`	SCImago Journal Rank: citations weighted by the prestige of the citing journals. Higher is better.
`sjr_quartile`	Q1 (top 25% of its subject category) to Q4 (bottom 25%).
`h_index`	The journal's h-index: h of its articles have each been cited at least h times.

The full data dictionary explains every column in every table, where each value comes from, and how to read the metrics.

Download the full data dictionary

A Markdown text file; it opens in any text editor.

Download the full dataset

Every university in one file per table. Each publication row carries the researcher's name, university and ORCID and the journal's name and ISSN, so it can be filtered or joined to another spreadsheet on its own. researcher_id and journal_id link the three files together; ORCID is the safest key for matching researchers to your own records, because names can collide and vary in spelling.

Publications Researchers Journals

CSV files; they open directly in Excel.

6. Known Limitations

Co-authorship counting. Publication counts are researcher–publication pairs. A paper co-authored by two staff members at the same university appears twice in the database — once for each author. Totals shown on the site reflect this.

Coverage depends on the university's own records. Papers missing from a university's system are only recovered if an open index (ORCID, Crossref or OpenAlex) has them. Work from before a researcher joined the university is often absent, particularly for Monash, where OpenAlex only adds papers written while at Monash.

Staff are chosen by department. As at every university, Adelaide staff are those whose profile lists the School of Accounting and Finance. A few researchers who work in accounting or finance but are based in another school, such as Management, are therefore not included.

Namesakes. Open indexes sometimes attach another person's papers to a researcher, especially for common names. The screening step removes the clear cases and lists them in <uni>_screened_out.csv, but a researcher whose genuine output is large can still carry some wrong papers. Corrections belong in the files in data/, not in the outputs, which the next run overwrites.

Journal coverage gaps. Not every journal has an ABDC rating, a Scimago entry, or a JCR Impact Factor. Null values in those columns indicate the journal was not found in the respective list, not that the journal is unranked.

Citation figures change between refreshes. Citation counts, FWCI and citation percentiles are live values from OpenAlex, so they drift every time the data is refreshed.

Academic levels. A small number of staff have no academic level, where the university lists an administrative title (such as "Deputy Head of School") or none at all.

7. Worked Example

A researcher at Monash University publishes a paper in the Journal of Finance (ISSN 0022-1082). Here is how the record is assembled:

Collect. The Monash adapter finds the researcher in the Accounting staff directory, reads their ORCID iD from their Monash research profile, and finds the paper in their profile's publication feed.
Identifiers. OpenAlex is searched by the researcher's ORCID iD, using only records that carry that exact ORCID. The OpenAlex copy with the same title supplies the paper's DOI and the journal's ISSN, 0022-1082.
Enrich. OpenAlex adds the citation count, FWCI, citation percentile and open-access status, looked up by DOI.
Rate the journal. ISSN 0022-1082 is found on the ABDC list as A*, in Clarivate JCR (impact_factor, impact_factor_5yr) and in Scimago (sjr, sjr_quartile, h_index, cites_per_doc_2y).
Screen. The paper came from Monash's own records, so the screening step leaves it alone.
Export and load. It is written to final output/monash/monash_publications.csv, with the journal in monash_journals.csv, and load.py puts both into the database.
The website displays the publication with an A* badge and the metrics in the table columns.

8. Universities Covered

The database covers Accounting & Finance researchers at the eight Group of Eight Australian universities. Each card shows the name stored in the database and the university's output folder.

Adelaide Universityfinal output/adelaide/

Australian National Universityfinal output/anu/

University of Melbournefinal output/unimelb/

Monash Universityfinal output/monash/

UNSW Sydneyfinal output/unsw/

University of Queenslandfinal output/uq/

University of Sydneyfinal output/usyd/

University of Western Australiafinal output/uwa/

Scraping is limited to researchers listed under Accounting & Finance on each university's business school research portal.