CLANCY ISR Fusion & Threat Assessment — UNCLASSIFIED // TRAINING USE ONLY

Ask a question

Try: probability of incursion · who is connected to SERPENT · threat level at Narva · track the S-400

Ask a question above.

Map

Toggles between the network view and the high-resolution satellite map. Click any point/node for its assessment.

Load injects

Accepts .json .kml .csv .xml .txt and .pdf. PDF text is extracted and split into injects on a best-effort basis (selectable-text PDFs only — scanned/image PDFs need OCR); review the result. “Clear imported” keeps the original seed TACREPs; “Clear all” empties everything for a blank slate. Both are undoable.

Add a report (image / video / audio / text)

Attach an image, video or audio file and click Analyse media to auto-fill a draft report, or type one directly. Image/video/audio analysis is simulated in this demo — a live system needs a vision / speech-to-text / NLP backend. Adding a report updates the network, threat and forecast (and is undoable).

Legend–

Nodes (by type)

Report (TACREP)

Proxy / criminal / info group

Call-sign

Military unit / air-defence

Location / hub

Vignette (grouping)

Links (co-mention strength)

Strong association

Weaker association

Low-credibility association

False / diversion — shown, 0 weight, ungrouped

Estimated location (inferred guess — no place named)

Bigger / brighter links = more corroborating reports. Magenta = flagged deception (visible but excluded from grouping & decisions). Click any node or link to inspect it.

Event probability (transparent)

No group selected.

Inspector

Click any node for a plain-English assessment.

What happens next

Building forecast…

The program's best estimate of the most likely next development, from the same model as the event panel. Click an entity for an entity-specific forecast; click a report or empty space for the overall picture.

Feed summary

Reports (TACREPs)–

Identifiable figures–

Associations–

Deception/false-leads–

Built from the CLANCY KML inject feed (54 TACREPs, 20 Feb–22 Mar 25). Identifiable figures are the primary connectors; time-distance is used as confirmation/projection.

Debug · diagnostics

Reports (n)–

Naïve pairwise (n²/2)–

Optimised comparisons–

Reduction vs naïve–

Spatio-temporal cells–

Proximity confirmations–

MENTIONS edges–

ASSOCIATED edges–

Graph build time–

The reduction applies to the time-distance confirmation layer (proximity/movement checks), where brute force is O(n²). It grows more dramatic as you import more reports. Primary linking is by identifiable figures (co-mention), which is already sub-quadratic.