AIR framework
Pathway and evidence terms
- Algorithmic Immersion Radicalisation (AIR)
- A developing socio-technical framework for investigating whether documented digital curation, sustained online immersion, reinforcement and interaction form part of pathways towards ideological mobilisation and harm. The label does not itself establish that an algorithm caused an offence.
- Algorithmic curation
- Automated ranking, recommendation or selection of content. Its presence on a platform is not evidence that a particular user saw, accepted or acted upon a specific recommendation.
- Immersion
- Sustained engagement with a digital environment, community or content stream. In this project it is an evidential domain, not a clinical state.
- Reinforcement
- Repeated social, informational or identity-confirming feedback that may strengthen attention or commitment. The term describes a proposed mechanism requiring evidence; it does not imply neurological measurement.
- Mobilisation
- Movement from belief, grievance or affiliation towards qualifying preparation, coordination, coercive conduct or violence. The project does not treat unpopular belief or lawful activism as mobilisation to violence.
- AIR level
- An expert-coded classification of the strength and type of documented public evidence for the project’s digital-pathway indicators. Higher, Moderate, Lower and Nil are evidence categories—not validated risk scores or severity rankings.
Case classifications
Recorded incidents and fields
- Terroristic attack
- An executed incident meeting the project’s operational conduct, purpose and broader-audience fear or coercion test. Legal classification and the project’s analytical classification are recorded separately where they differ.
- Thwarted terrorist plot
- Qualifying preparatory or attempted conduct disrupted before the intended terroristic attack was completed. Mere speech, aspiration or an uncorroborated allegation is insufficient.
- Coercive Ideological Violence Mobilisation (CIVM)
- Qualifying coercive, rights-infringing or violent ideological mobilisation that does not meet the project’s terrorism threshold. Lawful protest and protected expression are outside the category.
- Calculation included
- A retained record admitted to the analytical denominator under the current release rules. Exclusion from calculation may reflect unresolved eligibility, evidence or duplication; it is not a finding about guilt.
- Online mobilisation evidence
- Case-specific public evidence concerning digital communication, content use, production, coordination, community participation or other governed online-pathway indicators. General internet availability is not case-specific evidence.
- Primary target
- The single direct victim or target group selected under the project’s hierarchy. Symbolic interpretation does not replace the direct victim group unless the rule and evidence support it.
Country-year context
Indicator and comparability terms
- Social-context indicator
- A country-year measure describing part of the national technological, institutional, demographic, economic, health, housing, justice or social environment. It does not measure an individual offender’s exposure.
- Comparable with caveats
- A comparison is allowed, but a visible qualification concerning definition, source regime, population, period or measurement must accompany it.
- Within-country trend only
- The measure may support change-over-time analysis within a country but not direct level comparison between countries.
- Snapshot or context only
- The observation may provide descriptive context but lacks sufficient longitudinal or cross-country comparability for exploratory association.
- Series break
- A documented change in definition, instrument, population, method or source regime that limits comparison across the boundary. The visual preserves the break rather than drawing an uninterrupted line.
- Five Eyes mean
- An unweighted same-year mean displayed only when all five countries have compatible observations under the indicator’s permissions. It is not a population-weighted regional estimate.
Evidence and interpretation
How analytical statements are labelled
- Observed finding
- A descriptive result directly reproducible from an identified data release and denominator.
- Statistical association
- A relationship supported by a stated statistical analysis. Association does not, by itself, establish causation.
- Informed hypothesis
- A theoretically grounded AIR interpretation proposed for further testing, not a causally established conclusion.
- Methodological or contextual observation
- A pattern concerning sources, classification, reporting, law or the socio-technical environment that affects interpretation.
- Baseline index
- A transformation setting a valid selected baseline to 100 so differently scaled series can be compared by proportional movement. Indexing does not make incompatible underlying measures substantively equivalent.
- Exploratory lag
- A transparent shift of one annual series relative to another for hypothesis generation. A lagged visual pattern is not evidence of a causal delay.
- Missing
- No usable value is available under the release rule. Missing is not zero.
- Suppressed
- A value intentionally withheld because of disclosure, small-cell or interpretive safeguards. Suppression is not missingness and is not evidence of absence.
- Partial year
- A year for which the observation or case-collection period is not complete. It should not be compared with full years without a prominent qualification.