Other pages in this series (34)
Identification rate of comet and guest-star articles
Share of comet-keyword article dates that fall within ±90 days of a known comet's perihelion (JPL Small-Body Database, 376 comets with perihelion before 1911), against the share for the same number of random dates (1,000 draws).
Matching comet and guest-star report dates to known perihelia within ±90 days captures 78% of Sillok observation reports and 33% of Goryeosa articles, against 17% and 3% for the same number of random dates. The chance rate is the share of a source's span covered by ±90-day windows around known perihelia, so it differs by period; for the Ilgi (1623–1910) it reaches 33%, so read it with the sample precision (0.39).
- Formula
- Rate = articles inside the window / dated articles. The ±90-day window was fixed before computing and not adjusted. Control: random dates drawn from the source's span, as many as articles, 1,000 times (seed 20261102).
- How to read it
- The excess over the chance rate is the information. Sillok observation reports (the final 'yes' of the 1.0 precision sample — 64 coder yes plus 5 editorial overrides, 69 in all; up to 1.2.1 only the 64 were used), all keyword articles, the Goryeosa treatise and the Ilgi are shown side by side. The chance rate is the share of the source's span covered by ±90-day windows around known perihelia, so it rises for later periods with more known comets.
- Caveats
- Most historical orbits in the JPL list were themselves computed from Chinese records, so the check is not fully independent. Ilgi comet-keyword precision is 0.39 on a 107-article sample — 妖星 and 彗孛 are often omen rhetoric.
- External check
- Halley apparitions: Yeomans, D. K. & Kiang, T. 1981, 'The long-term motion of comet Halley', MNRAS 197(3): 633–646, doi:10.1093/mnras/197.3.633 (reference only — the match on this page uses the JPL list and does not test the Halley returns separately); the 1604 supernova: 323 '客星' articles in the Sillok of Seonjo's reign, 137 of them in 1604–1605 (guest_star in keyword_hits_sillok_ext).
- Reliability card
- sampledSampled precision — share of keyword hits that mean what the metric says, measured on a coded stratified sample
Keyword precision: Sillok 0.68 (n=101); Ilgi stratified sample of 107: 0.39 (Wilson 95% 0.31–0.49). The 'observation reports' rows use only the sample's final yes articles (coder plus editorial overrides).
| Set | Articles | Within ±90 d | Rate | 95% lo | 95% hi | Chance mean | Chance 97.5% | Excess |
|---|---|---|---|---|---|---|---|---|
| goryeosa: all comet-keyword articles | 40 | 13 | 32.5% | 20.1% | 48.0% | 3.0% | 10.0% | 29.6% |
| goryeosa_jeolyo: all comet-keyword articles | 9 | 3 | 33.3% | 12.1% | 64.6% | 3.0% | 22.2% | 30.4% |
| sillok: all comet-keyword articles | 1,605 | 1,071 | 66.7% | 64.4% | 69.0% | 17.2% | 18.9% | 49.6% |
| sjw: all comet-keyword articles | 1,686 | 1,382 | 82.0% | 80.1% | 83.7% | 33.2% | 35.3% | 48.7% |
| sillok: precision-sample observation reports | 69 | 54 | 78.3% | 67.2% | 86.4% | 16.7% | 26.1% | 61.6% |
| sjw: sample observation reports | 42 | 35 | 83.3% | 69.4% | 91.7% | 31.8% | 45.2% | 51.5% |