Types and platforms
Taxonomy, platform, and profile analysis
Public-safe aggregate analysis from the PRD chart inventory.
Top insight this month
Lagos and X continue to drive the largest share of visible signals, while body shaming remains the leading public category.
Co-occurrence pattern
Threats and doxxing overlap often enough to stay visible as an escalation warning, but detailed cases remain internal.
Classifier uncertainty
Pidgin and code-switched records move to analyst review when model confidence drops below the prototype threshold.
Incidents by violence type
Sortable horizontal Nivo bar chart for RACE taxonomy categories.
Incidents by platform
Platform-specific incident volume based on aggregate data.
Platform x category comparison
Cross-tab of platform and category counts.
Category co-occurrence heatmap
Paired taxonomy signals in the same aggregate cluster.
Survivor / target profile breakdown
Profiles are used for aggregate analysis, not identity exposure.
Severity x platform heatmap
High-safety overview of severity distribution by source.
AI confidence and uncertainty panel
Classifier confidence buckets for analyst attention.
High confidence
360>= 0.80
Needs analyst review
1140.60-0.79
Low confidence
42< 0.60 or code-switched
Incident explorer
Public-safe sample rows with anonymized summaries only.