# Citation Audit — JPL_ASTRO_EARTH_08

**Candidate:** JPL_ASTRO_EARTH_08
**Audited file:** `dissertation.md` (Part I: References, 149 consolidated entries)
**Method:** Every DOI was resolved against `https://api.crossref.org/works/<doi>`. Every
non-DOI URL (NTRS, OpenAlex) was checked by HTTP GET, with NTRS records resolved through the
hardened `ntrs.nasa.gov/api/citations/<id>` API rather than the SPA route (the NTRS single-page
app returns HTTP 200 for every path and therefore cannot distinguish a live record from a dead
one). Any DOI that did not return HTTP 200 at Crossref was re-checked against DataCite
(`api.datacite.org/dois/<doi>`) and the `doi.org` resolver before any "dead" classification, so
that a registration-agency mismatch is never mistaken for a dead identifier.

**Guardrail applied:** A 403 or timeout is *unverifiable*, never dead. A Crossref 404 alone is
*not* proof of death when the DOI is registered with a different agency (DataCite, mEDRA, etc.);
only a confirmed 404/410 across the resolver chain, or an ARGOS near-miss analogy, is counted as
a real defect.

## Summary

| Metric | Count |
|---|---|
| Total references | 149 |
| Resolved (live) | 149 |
| Unverifiable (403 / timeout / network) | 0 |
| Flagged dead | 0 |
| — of which mistyped_recoverable | 0 |
| — of which misattributed | 0 |
| — of which fabricated_neighbor | 0 |
| — of which not_found | 0 |

**Result: clean.** Every one of the 149 consolidated references resolves to a live,
correctly-registered source. No DOI is dead, no URL is broken, and no entry required the ARGOS
near-miss resolver (which is invoked only on confirmed-dead DOIs). There is no repair worklist:
nothing in this reference list needs to be corrected, replaced, or dropped.

## Notes on the two Crossref-404 entries (verified live, NOT defects)

Two DOIs returned HTTP 404 at Crossref. Both were confirmed live through the resolver chain and
are correctly classified as **resolved**, because their DOIs are registered with DataCite rather
than Crossref. Recording them here for transparency so a future reviewer does not re-flag them.

| Ref | Author-year / title | Cited DOI | Crossref | DataCite | doi.org landing | Disposition |
|---|---|---|---|---|---|---|
| 99 | "OPUS: An Integrated Assessment Model for Satellites and Orbital Debris (2023)" | `10.48550/arxiv.2309.10252` | 404 | 200 | `https://arxiv.org/abs/2309.10252` (HTTP 200) | **Live.** arXiv DOIs are minted through DataCite, not Crossref; the Crossref 404 is expected and not a defect. No fix needed. |
| 148 | Zhou et al., "In-hospital mortality from healthcare-associated infection by multidrug-resistant Pseudomonas aeruginosa..." *GMS Hygiene and Infection Control*, 2025 | `10.3205/dgkh000597` | 404 | 200 | `https://journals.publisso.de/.../dgkh000597` (HTTP 200) | **Live.** German Medical Science (publisso.de) DOIs register with DataCite; Crossref 404 is expected. No fix needed. |

## Parsing note (not a citation defect)

Reference 63 (Gelber & Gray, 1991, `10.1016/0197-2456(91)90089-5`) contains literal parentheses
inside the DOI suffix. A first-pass extractor truncated the DOI at the inner `(91`, which produced
a spurious 404. With the full paren-balanced DOI the entry resolves at Crossref with HTTP 200.
The dissertation's stored DOI is correct; the truncation was an audit-tool artifact only and has
been corrected in this report. No change to the reference is required.

## Flagged-reference table

None. No reference is dead, mistyped, misattributed, or fabricated. The ARGOS near-miss resolver
was therefore not invoked on any entry.