How to check the evidence behind each incident, where the news comes from, and how close the machine and the public come to institution decisions.
How close the machine (D1) and the public survey (D2) come to institution statements (D3) and official records (D4). The pipeline recomputes these numbers daily from each flow's latest snapshot.
computed 7 Oct, 08:19 WIB| Happened | Hoax | Not yet certain | Same as D4 | |
|---|---|---|---|---|
| D1 · Machine | – | – | – | –n=0 |
| D2 · Public survey | – | – | – | –n=0 |
| D3 · Institution | – | – | – | –n=0 |
| D4 · Official record | – | – | – | – |
| Field | Precision | Recall | F1 | n |
|---|---|---|---|---|
| Attack type | 1.00 | 1.00 | 1.00 | n=8 |
| Victim | 0.00 | 0.00 | – | n=1 |
| Actor | 1.00 | 1.00 | 1.00 | n=1 |
| Event date | 1.00 | 1.00 | 1.00 | n=2 |
| Location | 1.00 | 1.00 | 1.00 | n=5 |
| Field | D2 · Public survey | D3 · Institution |
|---|---|---|
| Attack type | 1.00n=3 | –n=0 |
| Victim | 0.00n=1 | 1.00n=1 |
| Actor | 1.00n=1 | –n=0 |
| Event date | 1.00n=2 | –n=0 |
| Location | 1.00n=3 | –n=0 |
| Field | Agree | Kappa | n |
|---|---|---|---|
| Attack type | – | – | n=0 |
| Victim | 0.0% | 0.00 | n=1 |
| Actor | – | – | n=0 |
| Event date | – | – | n=0 |
| Location | – | – | n=0 |
| Median hours | n | |
|---|---|---|
| D1 · Machine | 13.8 | n=2,053 |
| D2 · Public survey | – | n=0 |
| D3 · Institution | – | n=0 |
| D4 · Official record | – | n=0 |
| n | ||
|---|---|---|
| Refuted by institutions (D4) | 7 | n=18 |
| Passed the machine (high/medium trust) | 3 | n=7 |
| Passed the public (survey matches) | 4 | n=7 |
| n | ||
|---|---|---|
| Overlap of D1 and D4 steps (Jaccard) | – | n=0 |
| D4 records adding steps beyond D1 | – | n=0 |
| Overlap of D1 and D4 literacy audiences (Jaccard) | – | n=0 |
| Overlap of D1 and D4 literacy topics (Jaccard) | – | n=0 |
| All incidents | Reached D4 | |
|---|---|---|
| high trust | 1.2% | 6.7% |
| medium trust | 7.8% | 26.7% |
| low trust | 91.0% | 66.7% |
| not scored | 0.0% | 0.0% |
– means no data yet; n is the number of incidents or fields counted. See the per-incident comparison on each incident's detail page.