Active Learning for Road-Network Conflation Ground-Truth Labelling: Pseudocode and Labelled Corpora (Niigata KSJ, Policies A/B/C)

Companion artefact for the manuscript "Active Learning for Road-Network Conflation Ground-Truth Labelling: A Cross-Policy Study on the Niigata KSJ Cascade" (International Journal of Intelligent Transportation Systems Research, ITS Symposium Japan refereed track, manuscript IJIT-D-26-00503, under review). The deposit contains (1) algorithm-level pseudocode with flow diagrams for every computational component of the paper: the nine-criterion cascade matcher and five-pass pipeline, the six-dimensional feature vector and composite score, three calibrators (isotonic, Dirichlet two-stage, split-conformal), eight acquisition strategies, the active-learning simulation loop, the six-encoding bearing ablation, the full statistical protocol, and figure generation; and (2) three hand-labelled corpora from the Niigata three-city polygon (Policy A, direction-independent, N=500; Policy B, direction-strict, N=500; Policy C, geometry-only re-labelling, N=250), as CSV and Parquet with a schema README. Corpora are derived from "National Land Numerical Information (Road)" (Ministry of Land, Infrastructure, Transport and Tourism, Japan), redistributable under the MLIT Download Site Content Terms of Use; the modified content is not endorsed by the Ministry. TomTom identifiers appear as pair keys only; no probe data is included, and the upstream probe-segment table is not redistributable. This work was supported by the Ministry of Land, Infrastructure, Transport and Tourism's Small Business Innovation Research (SBIR) Phase 3 Fund Project.