Learned process groups expose different time profiles in traffic-fine records
Source check: Source checked. Not independent scientific confirmation.
Finding: NMF-based process grouping in150370 traffic-fine records produced groups concentrated in different periods despite calendar dates being excluded from their encodings. Three groups had88-94% of traces ending in2000-2004, while two others had86-97% ending in2009-2013. These exploratory temporal associations do not identify the causes of process change.
Technology: TracePrism; non-negative matrix factorization; k-means
Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Fitted topic representations and learned partitions supply behavior-based groups whose subsequently examined date distributions reveal the temporal contrast. UMAP is a visual aid rather than the enabling operation.
- Test and result
- An analyst iteratively fitted NMF topics and formed process groups, with k-means used on topic weights in a later round. Posthoc end-date distributions compared accepted group memberships.
- Limitation / still unknown
- Group thresholds, encoding choices and commit order were analyst-selected on the same log. No reseed stability study or dedicated drift analysis separated behavior from calendar policy changes; late open cases can reflect right-censoring. The result is a data-specific association, not evidence of an administrative cause or improved process performance.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://arxiv.org/abs/2610.28532v1 · https://arxiv.org/html/2610.28532v1
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
Review date: 21 Nov 2026. Review scope: Primary v1 front and safe computational/model Methods, specific application Results and statistical limitations.
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.