Citations
Which research fields reference one another? This is the connection measured by the score.
How connected is trust research?
Follow citations across disciplines. Explore the papers and the calculation behind the TRUST TRACE Connection Score.
Loading the local citation map…
Historical prototype; relevance screening pending. The recovery includes different meanings of “trust”, including technical and mathematical uses. They remain in this unscreened analysis. The score describes this collection and is not ready to represent the Foundation’s final research scope.
TRUST TRACE Connection Score
38.9/100
Of every 100 citation links mapped inside TRACE, 38.9 connect papers assigned to different research fields.
All fields · 2019–2023 · historical collection
31,751 ÷ 81,570 × 100
A measure of citation connections inside this collection. It does not establish the quality of research, integration of ideas or stakeholder participation.
Two kinds of connection.
Across fields38.9%Within a field61.1%
31,751 cross-field links among 81,570 mapped citations.
13.9% of 586,580 recorded references map to eligible papers inside TRACE.
9,176 of 11,339 eligible citing papers contribute at least one mapped link. Read the scope ↓
01 / Where fields meet
Follow a research connection.
Select a field pair to inspect the actual citation links. The number beside each pair counts links, not researchers or projects.
02 / Through time
Does the pattern change?
A fixed historical collection, grouped by the citing paper’s year. Select a year to examine its connections.
The vertical scale runs from 0 to 100. Years with fewer than 100 mapped links show a grey marker. Collection coverage and classification can affect the pattern; this chart does not establish progress in interdisciplinarity.
Read the annual values as a table
| Year | Score | Mapped links | Across fields |
|---|
03 / Check the evidence
Every connection has two papers.
The first paper cites the second paper. Open either title for its identifier and assigned field.
Title search and the relationship buttons filter this evidence list. The headline score continues to describe the selected period and citing field. Newest citing papers appear first.
04 / An open calculation
A score you can take apart.
One defined relationship and one denominator. Geographic collaboration and content-based assessments remain separate.
What enters the calculation?
- Select citing papers by publication period and primary field.
- Follow their recorded references to other eligible papers in the historical TRACE collection.
- Count a cross-field link when the two papers have different OpenAlex primary-topic fields.
- Divide cross-field links by all eligible mapped links and multiply by 100.
31,751 cross-field links ÷ 81,570 mapped links × 100
Each directed citation link counts once. A paper citing several other papers contributes several links. This describes the citation network, rather than an average rating of individual papers.
Comparison: if citations were matched at random while each field kept its share of citing and cited links, about 79 in 100 (78.5) would cross fields in 2019–2023. The observed 38.9 is well below that.
What does the score leave open?
A citation can connect fields without integrating their methods. Two papers in the same broad field can still represent different perspectives. This indicator makes one observable connection visible; it is not a validated scale of transdisciplinarity.
TRACE follows the Foundation’s ambition to connect knowledge across boundaries. Evidence of shared problem definition, knowledge integration and stakeholder participation belongs in separately reviewed framework annotations.
Explore the TRUST framework →Collection, dates and exclusions
The starting recovery contains 25,408 records. This score uses a fixed universe of 25,328 records dated 1990–2023 after excluding 58 cached retraction flags and 22 records from the incomplete 2024 slice. The default citing period is 2019–2023. Cited targets may be older. References dated later than the citing paper are withheld as date inconsistencies: 33 in the selected view. These rules create an analysis view; the underlying records remain preserved. Scope screening of the historical recovery is still pending.
Coverage and missing references
Only references with a matching eligible work in TRACE enter the score. References outside the collection, unresolved identifiers and absent reference lists are not treated as within-field links or as evidence of no connection. Mapping coverage and contributing-paper counts are shown beside the score. The 13.9% default mapping coverage describes this bounded network, not complete reference coverage in OpenAlex. A score for the whole literature would need a different, sufficiently complete collection.
Fields, sparse views and interpretation
Each paper uses one field from its cached primary topic. These broad categories are classification proxies, not an author’s declared discipline. A cross-field link is unweighted by intellectual distance or citation sentiment. Displayed percentages are withheld below 100 mapped links; the underlying counts remain available. This is a display rule, not a validated statistical threshold. There is no target, pass mark or claim that a higher score is inherently better. Values across different classification systems should not be compared directly.
Can we calculate this without full texts?
Yes. The score uses cached work identifiers, reference links and field assignments. No paper full text or abstract is needed for this calculation. Titles and available abstracts can later support reviewed trust categories. Author affiliations support country and organisation indicators; they do not establish a person’s discipline or stakeholder contribution. The broad cached recovery did not retain abstracts, and the new discovery sample is not representative enough to substitute for it.
05 / Test the choice
How much does the method matter?
These checks hold the default 2019–2023, all-fields view fixed. They compare different definitions and scopes; the range is not a confidence interval.
| Calculation | Score /100 |
|---|---|
| Selected definition | 38.9 |
| Retain date-inconsistent references | 38.9 |
| Equal weight per contributing paper | 38.1 |
| Both primary-topic scores at least 0.5 | 37.1 |
| Omit Computer Science at either end (scope stress test) | 36.5 |
The primary-topic cutoff is a sensitivity test, not a validated probability threshold. The Computer Science omission tests dependence on a large field; it is not an approved human-versus-technical trust classifier.
06 / Beyond a single number
Three kinds of evidence. Three different questions.
Titles and abstracts
Who trusts whom, in which situation, and with what risk? Reviewed annotations can connect the literature to the TRUST framework, with unknowns kept visible.
Author affiliations
Where are the contributing organisations, and which kinds participate? Geography and organisational mix add context without being mistaken for disciplinary integration.