Declare the incident reference set
Synthetic confirmed incidents are I1,I2,I3,I4. Notifications reference I1 three times, I2 twice and I3 once; none references I4. Every notification is correctly associated, and no other incident exists. A duplicate is any message beyond the first for the same incident in this review.
| Incident | Messages | Extra after first |
|---|---|---|
| I1 | 3 | 2 |
| I2 | 2 | 1 |
| I3 | 1 | 0 |
| I4 | 0 | 0; missed |
Match notifications to identities
- Group messages by incident
- Count covered incident IDs
- Count extras and missing IDs
A message with no resolved incident identity remains unclassified, not a duplicate by guess.
What are coverage and duplicate-message share?
Show the worked answer
Coverage 3/4=75%; extras 3 of 6 messages=50%. I4 is the missed incident. Removing extras does not fill that missing coverage.
Report noise and misses separately
| Case | Decision | Reason |
|---|---|---|
| Reduce 6 messages to 3 | Noise reduced | Same 3 incidents covered |
| Add first I4 notification | Coverage becomes 4/4 | Missing incident covered |
Use separate review actions for I4 detection and repeated-message handling. A quieter inbox alone does not establish better incident coverage.