# Citation Audit, JPL_MGMT_SMA_TECH_04

Read-only audit of the dissertation reference list. The dissertation file was NOT modified. This document is a repair worklist for the candidate's reference section.

## Summary

- **Total references:** 116
- **Resolved (live):** 113  (103 DOIs via Crossref 200, plus 10 URLs/NTRS reachable)
- **Unverifiable (NOT defects):** 2  (DataCite DOI absent from Crossref REST but live via doi.org; publisher bot-block)
- **Flagged dead (confirmed 404/410):** 1

### Counts by classification (flagged items only)

| Classification | Count |
|---|---|
| mistyped_recoverable | 0 |
| misattributed | 0 |
| fabricated_neighbor / not_found | 0 |
| dead link (real source, wrong URL) | 1 |
| unverifiable (resolves elsewhere / bot-block) | 2 |

**Bottom line:** Every cited source in this reference list corresponds to a real, locatable publication. There are **zero fabricated DOIs**, zero mistyped DOIs, and zero misattributed DOIs. The single genuine defect is one broken landing-page URL (ref 59) that points to the wrong conference-session subdirectory for a paper that does exist. Two further entries are flagged only because Crossref cannot confirm them (ref 96 is a DataCite DOI; ref 114 is behind a CNKI bot-block); both are live and per the guardrail are NOT defects.

### Guardrail note (two first-pass false positives cleared)

- **Ref 51** (`10.1016/s0951-8320(99)00037-x`) first returned 404 because the audit script URL-encoded the literal parentheses in the DOI, which the Crossref `/works/` endpoint mishandles for legacy Elsevier DOIs. With the raw path it returns 200 and doi.org resolves it to the Elsevier article. **Real, resolved, no defect.**
- **Ref 96** (`10.7275/7tgr-xt91`) returns 404 from the Crossref REST API because `10.7275` is a **DataCite** prefix (UMass ScholarWorks / openpublishing), not a Crossref-registered DOI. doi.org resolves it to the live article at openpublishing.library.umass.edu/pare/article/id/1591. Per the guardrail (a DOI that resolves via doi.org is not dead), this is **unverifiable-via-Crossref only**, not a defect.

## Flagged references (repair worklist)

| Ref | Author-year (short) | Cited DOI / URL | Classification | Suggested fix |
|---|---|---|---|---|
| 59 | M. Langer and J. Bouwmeester, "Reliability of CubeSats - Statistical Data, Developers' ... | https://digitalcommons.usu.edu/smallsat/2016/S3GuidCont/2/ | **dead URL (real source, wrong path)** | Replace with the correct USU DigitalCommons landing page for Langer & Bouwmeester, "Reliability of CubeSats," *30th AIAA/USU Conf. Small Satellites*, 2016. The cited `/smallsat/2016/S3GuidCont/2/` path returns 404; the conference root `/smallsat/2016/` is live. Locate the paper's actual session entry (or cite via its SSC16 paper number) and update the URL. |
| 96 | P. Scott, "Causal Inference Methods for Selection on Observed and Unobserved Factors: P... | `10.7275/7tgr-xt91` | unverifiable (DataCite, not in Crossref) | No change required for correctness: `https://doi.org/10.7275/7tgr-xt91` resolves to the live UMass article. Optional: note it is a DataCite-registered DOI so future Crossref-only checks do not re-flag it. |
| 114 | Z. Yilin, "A Study of Influence of Technology Readiness Level on Spacecraft Development... | https://en.cnki.com.cn/Article_en/CJFDTOTAL-HTGC200902004.htm | unverifiable (publisher bot-block) | No change required: CNKI returned HTTP 418 to the automated checker (bot-block), not a 404. The CNKI record for Zhao Yilin, "Influence of TRL on Spacecraft Development Schedule" (*Spacecraft Engineering*, 2009) exists. Optionally add a DOI if one is available, or confirm the URL in a browser. |

## ARGOS resolver evidence (dead-DOI characterization)

The ARGOS near-miss resolver was run on every DOI that did not resolve on the first Crossref pass (refs 51 and 96). Quoted evidence:

**Ref 51**, resolver verdict `resolved`:

> "the cited DOI resolves; no resolution-grounds defect" ... exact.resolves=true, title "Weibull and inverse Weibull mixture models allowing negative weights," *Reliability Engineering & System Safety*, 1999. Confirms the first-pass 404 was an encoding artifact.

**Ref 96**, resolver verdict `misattributed` (Crossref-only view):

> "the cited DOI does not resolve, but real papers on this topic exist under different DOIs (e.g. 'Propensity score matching and causal inference', 10.1920/ps.ifs.2024.1048) ... the candidate cited a non-existent identifier for a real literature and must correct or replace it"

This ARGOS verdict is **superseded** by direct doi.org resolution: `10.7275/7tgr-xt91` is a DataCite DOI that the resolver's Crossref-only probe cannot see, yet it resolves live to the UMass ScholarWorks article that is exactly the cited Scott (2019) source. The resolver's topic-neighbor suggestions are therefore not needed; the cited identifier is real, just outside Crossref. Recorded here for transparency; no replacement is warranted.

## Method

1. Extracted all 116 `<span id="ref-N">` entries (104 with DOIs, 12 URL-only).
2. Resolved each DOI against `https://api.crossref.org/works/<doi>` and recorded the HTTP code; classified 200=resolves, 404/410=dead, 403/429/timeout=unverifiable.
3. Checked the 12 non-DOI URLs with a GET; NTRS citations were checked via `ntrs.nasa.gov/api/citations/<id>` per the hardened rule (the NTRS SPA 200s every route).
4. Ran the ARGOS near-miss resolver on every non-resolving DOI.
5. Re-verified every flagged item via doi.org (not just Crossref) before finalizing, applying the guardrail that a 403/timeout or a DataCite-only DOI is unverifiable, never dead.