Pattern Matching Across USDA Inspection Records

Once inspection reports and published OIG products are held in one place, the useful next step is comparison. The same 9 CFR section, similar species or facility type, a shared region, an inspector’s recurring phrasing, or a theme that also appears in an OIG audit can be set beside each other. That work is pattern matching. It is not a finding that one event caused another, and it is not a USDA determination.

Why comparison requires a held corpus

APHIS’s Public Search Tool is built to look up facilities, inspections, and related compliance records.¹² It is not built to answer “how often has this citation language appeared for this class of exhibitor since 2018?” That question needs fields the portal does not join: section number, species category, site type, date, and the narrative text itself. A warehouse that stores those fields—and the original PDFs—can run the comparison. A fresh portal search cannot.

The comparison is only as good as the files held. If the warehouse is missing a year, a certificate series, or an OIG report that bears on the same theme, the pattern is a pattern in an incomplete set. The method note has to say so.

What a defensible match looks like

A match should name the dimension being compared and the sources used. Typical dimensions in this record are:³

• The same 9 CFR section or subsection.
• Comparable species standards (for example, Part 3, Subpart A versus Subpart F).
• Facility category: dealer, exhibitor, research registrant.
• Geography or inspection year.
• Recurring narrative phrases that are not themselves citations.
• An OIG audit theme that concerns the same Animal Care control, not the same private party.

Each match then needs two attachments: the inspection or OIG files that contain it, and a one-line statement of what was not compared. “We compared posted inspection reports in the held set for certificate type Class C from 2019 through 2024” is a method. “Inspectors always cite this” is not.

What recurrence does not show

APHIS’s inspection guide states that it cannot replace an inspector’s professional judgment and that it does not change the regulations.⁴ Frequency of a citation, therefore, can reflect how often a condition is documented. It can also reflect inspection intensity, a change in how a section is applied, or a change in what is posted. The public file rarely contains the denominator—how many similar facilities were not cited.

OIG work has the same caution at a different scale. Audit Report 33601-0002-31 examined APHIS controls over dog-breeder oversight and recorded recommendations about data reliability and complaint response. It did not purport to adjudicate individual licensees.⁵ OIG’s later management-challenges compilation tracks whether those recommendations remain open.⁶ Using an OIG audit as if it were a citation against a named business is a category error. Using it as context for how Animal Care’s own controls were reviewed is appropriate.

Software can retrieve. It cannot stand in for the exhibit.

Search and matching may be assisted by software. The Inspector General Act’s audit-and-investigation duties, and APHIS’s posted inspection series, remain human-readable public records.⁷² A generated summary that drops the file, the section number, or the date is not a match. The packet should still open to the same PDFs a second researcher would use.

That is the standard this work is being built to meet: a hosted warehouse of the public USDA record; queries against what is held; patterns that are labeled as patterns; and an evidence packet that keeps the sources attached.