Community WolfSIA
Research

Grounded in research. Configurable by design.

SIA operationalises decades of established crime-and-place theory as configurable features, then exposes every assumption for expert review. The intent is not to replace experts. It’s to give them a calibration foundation that didn’t exist before.

§ 01 · Academic foundation

Six bodies of work. One inspectable foundation.

These are the bodies of theory the platform consciously operationalises, and the platform features that carry them. Where the live score only approximates a theory, we say so. Where a theory is on the roadmap rather than shipping, we say so too.

TheoryKey insightHow SIA represents itStatus
Routine Activity Theory
Cohen & Felson (1979)
Crime occurs when a motivated offender, a suitable target and the absence of a capable guardian converge in space and time.Crime feeds + infrastructure attractors/generators + guardianship places (defenders) + survey-derived security visibility.Strong
Crime Pattern Theory
Brantingham & Brantingham (1995)
Crime concentrates around nodes, paths, edges, generators and attractors.Transit, malls, nightlife, taxi ranks, dense built-environment clusters, highway junctions: encoded in the Infrastructure Ontology.Strong
Hot Spots & Micro-Place Research
Braga et al. (2019); Weisburd
A small number of places are responsible for a disproportionate share of crime.H3 resolution 8 hex grid; forensic per-hex details endpoint; banding tuned to micro-place concentration.Strong
CPTED & Environmental Criminology
Multiple systematic reviews
Lighting, visibility, disorder and access control shape opportunity and fear.Street lighting features, abandoned-building survey factors, sightlines, broken-windows perception scoring.Configurable
Crime Harm Indices
Sherman et al., Cambridge CHI (2016)
Crimes must be weighted by harm and severity, not raw counts.Configurable type weights, intent relevance, severe-crime parallel curve. Approximated today; formal indexing on the Phase-2 roadmap.Phase 2
Social Disorganisation Theory
Shaw & McKay (1942); Sampson et al. (1997)
Concentrated disadvantage, residential mobility and collective efficacy drive violence.Recognised but not yet fused into the live score. Phase-2 priority for socioeconomic and deprivation layers.Phase 2
§ 02 · In-house domain reviewer

An expert in the room, not in the footnotes.

Most vendors cite criminology. We employ it. Phase-1 narrative, default scoring posture and ontology framing are reviewed against the literature they claim to operationalise.

Spatial crime researcher

Dr Gretha Groeneveld · PhD spatial crime research

Published cross-national spatial-crime work using ISO 37120 indicators to operationalise social-disorganisation theory across Khayelitsha (South Africa) and Fort Lauderdale (USA) with G. D. Breetzke (2023, International Criminal Justice Review). Active collaborator on Community Wolf in the Stellenbosch context. Reviews the Phase-1 narrative, default model posture and ontology framing on this platform.

Routine Activity TheoryCrime Pattern TheoryRational ChoiceHot Spots · Micro-placeNear-repeat VictimisationCPTEDHarm-weighted MeasurementSocial DisorganisationCollective Efficacy
§ 03 · Tunable models, calibrated by experts

Every assumption is reachable from a single surface.

Crime-type weights, built-environment risk classes, report and survey factors, source budgets, intent presets: every assumption the score depends on is reachable from one surface. Researchers and customers can author named, versioned tunable models (your model, your weights) and switch between them atomically against a tenant.

That’s the practical answer to a deep methodological critique: no single criminologist’s opinion is hard-coded. The platform makes the choice itself the product surface.

§ 04 · What we say · what we won’t

Honest framing, published.

These are the rules we hold ourselves to in external communication. We publish them because procurement, counsel and academic reviewers benefit from a vendor that commits, in writing, to what it won’t claim.

We say this
  • Criminology-informed prioritisation score built on established crime-and-place theory.
  • A transparent, availability-aware Risk API composed from four signal sources.
  • Customers and researchers can save their own tunable models (your model, your weights) on top of our data.
  • Reviewed in-house by Dr Gretha Groeneveld, PhD spatial crime research.
We won’t say
  • Peer-reviewed predictive model (the in-house forecasting capability is internally validated, not externally peer-reviewed on this build).
  • Implements the Cambridge Crime Harm Index (we approximate harm weighting; formal indexing is roadmap).
  • Includes social-disorganisation variables in the live score (it's a Phase-2 fusion priority, not shipped).
  • The most accurate safety score - absolutes belong to vendors, not infrastructure.
§ 05 · Toward research-validated

Phase-2 priorities: where capital converts to defensibility.

  • Formal harm indexing: separating raw incident counts from severity-weighted measurement, in line with the Cambridge Crime Harm Index and the broader harm-index literature.
  • Socioeconomic and deprivation fusion: explicit fusion design tied to social-disorganisation literature; never precinct rebadged as census sociology.
  • Spatial holdouts and ablation studies: published methods, calibration curves by crime type, infrastructure channel ablation, cross-checks against open official statistics.
  • Pre-registered partner studies: hotspot and walking outcomes; move from research-aligned to research-validated.
  • Predictive integration: retrain the in-house GCN against the production hex graph and ship as POST /forecast/hex through the same platform contract.
§ 06 · Further reading

The literature lineage: bibliography on request.

A full reference bibliography accompanies investor and academic conversations as a leave-behind document, nineteen sources spanning Cohen & Felson’s Routine Activity Theory through the Cambridge Crime Harm Index. Every theoretical claim on this page is traceable to its source. Email research@ communitywolf.com to request the leave-behind, or open an enterprise conversation below.