1-year window
Emphasizes the most recent citation response.
Methodology
How journal and author evidence becomes a ranking view.
The method is split by entity. Journal indicators measure citation evidence across explicit windows; author indicators estimate contribution across linked publications and citations. Raw values remain visible wherever possible.
Coverage and methods are continuously updated as new evidence and corrections are verified. Submit supporting evidence to admin@scientificworthindex.com.
Scope and terminology
Scientific Worth Index uses three public evidence layers. Record is the broad monitored registry, Index is the evaluated middle layer, and Core is the strongest visibility layer. These labels organize the product; they are not claims of official inclusion in another commercial index.
Broad layerRecord
Evaluated layerIndex
Highest layerCore
Visible groupingQuartile Q1-Q4
Journal indicators
Journal profiles distinguish total citations, external citations and self-citations. Index-1-Year, Index-2-Years and Index-5-Years use the corresponding prior publication windows. Eligible external citations are associated with the cited publication year and normalized against journal output in the same window.
2-year window
Balances recency with a broader publication base and supports current quartile placement.
5-year window
Shows a longer citation horizon for fields with slower accumulation.
Self-citation is reported separately. External cited names requiring review can remain auditable without receiving an external quartile assignment until classification is resolved.
Quartile classification
The public label is Quartile: Q1, Q2, Q3 or Q4. Historical data fields may still use the internal name quarter, but the visible concept is quartile placement. Quartile weights used by the current model are Q1 = 1.00, Q2 = 0.75, Q3 = 0.50 and Q4 = 0.25.
A quartile is meaningful only with its year and method. Profiles therefore show annual rows and avoid extending charts across years where no evidence exists.
Worked 2-year Index example
Assume a journal receives eligible external citations in 2026 to articles it published in 2024 and 2025. The citing journals contribute 120 citations from Q1, 80 from Q2, 40 from Q3 and 20 from Q4. SWI applies the previous-year quartile of each citing journal.
| Citing-journal quartile | Eligible citations | Weight | Weighted contribution |
|---|---|---|---|
| Q1 | 120 | 1.00 | 120 |
| Q2 | 80 | 0.75 | 60 |
| Q3 | 40 | 0.50 | 20 |
| Q4 | 20 | 0.25 | 5 |
Weighted external score: 120 + 60 + 20 + 5 = 205. If the journal published 50 eligible articles in the same 2-year publication window, its Index-2-Years value is 205 / 50 = 4.10. Citations with an unknown cited-publication year do not enter the window score.
Journal Quality and Undeserved Authorship Risk
These two indicators summarize metadata and evidence signals attached to articles published by a journal in a given year. They do not measure the scientific truth, novelty or peer-review quality of the research. A missing article-level score is excluded from the annual average rather than converted into zero.
Journal Quality: definition and calculation
Each scored article receives a value from 0 to 100. The current model combines five components:
| Component | Weight | Current input logic |
|---|---|---|
| Metadata completeness | 30% | Share of seven fields present: title, DOI, abstract, year, language, journal identity and source URL. |
| Reference depth | 25% | Reference count divided by 30 and capped at 1.00. |
| Journal match rate | 20% | Share of references that can be linked to a journal in the current evidence graph. |
| PDF availability | 10% | Component value 1.00 when a PDF path or URL is available; otherwise 0.40 in the current heuristic. |
| Openness signals | 15% | Open-access metadata contributes 0.40 of this component; ethics-statement and data-availability signals contribute 0.30 each. |
Article Quality = 100 x (0.30 x metadata completeness + 0.25 x reference depth + 0.20 x journal match rate + 0.10 x PDF availability + 0.15 x openness signals). The journal-year value displayed in profiles is the arithmetic mean of available article scores for that journal and year.
A higher value means that the covered articles contain more of the metadata, reference-linking and access evidence used by this model. It must not be read as a verdict on research merit, editorial standards or the standing of individual authors. Coverage practices differ across publishers, years, document types and disciplines, so comparisons require those limitations to be considered.
UAR: definition and calculation
Undeserved Authorship Risk (UAR) is a bounded metadata heuristic from 0 to 1. For each article, the current model adds the following rule weights and caps the result at 1.00:
| Observed signal | Added risk weight |
|---|---|
| Eight or more listed authors | 0.35 |
| Five or more authors linked to no more than one institution | 0.25 |
| At least 30% of listed authors have no affiliation in the available metadata | 0.20 |
| The source metadata explicitly reports no ethics statement | 0.10 |
The journal-year UAR is the arithmetic mean of available article-level risk scores. Higher values mean that more of these rules were triggered in the available records. The profile chart multiplies UAR by 100 only to place the 0-1 series on the same visual scale as Journal Quality; the annual table retains the 0-1 value.
Misuse warning: UAR does not identify gift authorship, misconduct or an ethics violation. Large-team disciplines, consortium papers and incomplete affiliation feeds can raise the score without wrongdoing. The indicator is a review prompt, not an accusation, and is not currently normalized by discipline or document type.
Known limits and responsible use
- Name variants, incomplete ORCID coverage and changing affiliations can split or merge author records.
- Publisher and metadata coverage can differ by journal and year.
- Field-normalized author positions are not shown because the current author dataset does not yet contain a reliable subject field.
- Rankings should support investigation, not replace expert evaluation or peer review.
Material method changes and data updates should be read together with the update date shown on ranking and profile pages.