RSKGD-FL-2018-2019: A Two-Year Florida Temporal Road Safety Knowledge Graph Benchmark

RSKGD-FL-2018-2019 is an independent two-year Florida temporal road-safety knowledge graph benchmark. It provides harmonized crash-event, roadway-reference, weather-context, edge-month, and edge-year analytical tables; temporal knowledge-graph node and relation partitions; schemas; data dictionaries; manifests; provenance records; and benchmark task definitions for prospective year-shift, month-forward, hotspot-persistence, and temporal road-risk analyses. The dataset is designed for research on road-safety knowledge graphs, temporal road-risk memory, non-i.i.d. evaluation, hotspot persistence, and rare-event representation learning. It is not a new version of RSKGD-FL-2018. It is a separate dataset record that builds on and cites the earlier Florida 2018 release as a prior published dataset, source/provenance anchor, and schema reference. Prior dataset:ELBASRI, A. (2026). RSKGD-FL-2018: A Florida 2018 Road Safety Knowledge Graph Dataset (1.0.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.20854520 The present dataset should be cited separately as:ELBASRI, A. (2026). RSKGD-FL-2018-2019: A Two-Year Florida Temporal Road Safety Knowledge Graph Benchmark (1.0.0) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.21229915 Users should read README.md, DATASET_CARD.md, LICENSES_AND_TERMS.md, and the provenance and manifest files before analysis. Candidate roadway-reference links, weather context, and exposure-related fields are analytical/proxy layers and should not be interpreted as official agency products or ground-truth crash-road assignments.