Link and connect incidents
Summary: You will understand how the platform proposes connections between incidents, and how to assert or review a connection yourself.
How the platform proposes connections
Every incident is automatically compared against nearby incidents the moment it is saved, imported, or edited. When two incidents look similar enough, the platform links them as a proposed connection: a suggestion, not a conclusion. These connections feed the crime network picture, so a pattern of related incidents can surface across stores, dates, and cases before anyone spots it by hand.
The comparison weighs several kinds of similarity together: matching suspects and vehicles (a shared license plate counts as strong evidence on its own), the same products or method or incident type, and how close the incidents are in place and time. A pair must reach a minimum strength to be proposed at all; weaker pairs are recorded nowhere and nobody sees them.
A proposed connection waits in a review queue for an analyst to confirm or dismiss it. Confirming keeps it as part of the network picture; dismissing removes it. Editing the underlying incidents later re-checks the strength automatically, but a confirm-or-dismiss decision you already made is preserved: the platform does not silently re-open or re-close a call.
Add a connection yourself
You can assert a relationship between two incidents directly, without waiting for the platform to propose it.
- Choose a reason for the connection: same suspect, same product, same method, same location, close in place and time, detected similarity, or "other."
- If you choose other, you must add free-text notes: the connection is rejected without them.
A manual connection is treated as confirmed at full confidence, unlike a platform-proposed one, which waits for a decision.
Each pair of incidents can be connected only once (the platform ignores the two possible orderings), and an incident cannot be connected to itself.
Review similar incidents
The "similar events" view of an incident shows the incidents it has already been scored against: confirmed and proposed, with dismissed pairs left out.
- The strength shown for a pair is always the stored score, the same number wherever it appears.
- An incident that has never been compared (for example, one saved while comparison was briefly unavailable) shows nothing here. Re-saving the incident triggers a fresh comparison.
- Incidents that share a cluster (a platform-formed group of related incidents) with the one you are viewing are highlighted and ranked ahead of the others. You do not create or edit clusters; the platform forms them, and an incident can belong to more than one.
Good to know
- The review queue for platform-proposed connections is part of the intelligence tools, which are being introduced separately and are not covered in this edition.
- A bulk re-comparison of every incident exists as an administrator action, for cases where many incidents need to be re-scored at once.