sub:assertion {
  sub:dggs-benchmark-outcome-2026 <http://schema.org/endDate> "2026-03-07"^^xsd:date ;
    a <https://w3id.org/sciencelive/o/terms/FORRT-Replication-Outcome> ;
    rdfs:label "DGGS Benchmark Outcome: Claim Partially Supported" ;
    <https://w3id.org/sciencelive/o/terms/hasConclusionDescription> """The claim \"Effect of DGGS Indexing on Associating Vector and Raster
Geospatial Data\" is PARTIALLY SUPPORTED.

VECTOR BENCHMARK (Figure 6): VALIDATED
DGGS provides orders of magnitude performance improvement over traditional
vector overlay operations. At 20 layers, DGGS was 16,000x faster than
vector methods.

RASTER BENCHMARK (Figure 7): PARTIALLY SUPPORTED
The paper's claim of \"roughly equivalent performance\" holds when comparing
classification time with pre-indexed DGGS data. However, on-the-fly H3
indexing adds significant overhead. The replication using xdggs shows
vectorized indexing reduces this overhead by ~100x.""" ;
    <https://w3id.org/sciencelive/o/terms/hasConfidenceLevel> <https://w3id.org/sciencelive/o/terms/HighConfidence> ;
    <https://w3id.org/sciencelive/o/terms/hasEvidenceDescription> """VECTOR BENCHMARK RESULTS:
| Layers | DGGS     | Vector   | Speedup  |
|--------|----------|----------|----------|
| 5      | 0.01s    | 0.4s     | 40x      |
| 10     | 0.015s   | 10s      | 670x     |
| 20     | 0.03s    | 400s     | 16,000x  |

DGGS shows near-linear scaling; vector shows super-linear growth.
This validates the paper's Figure 6.

RASTER BENCHMARK RESULTS (100 layers):
| Method              | Time    |
|---------------------|---------|
| Raster (NumPy)      | 0.02s   |
| DGGS Pre-indexed    | 0.01s   | ← Paper's scenario: VALIDATED
| DGGS + H3 loop      | 5.0s    | ← Includes slow indexing
| DGGS + xdggs        | 0.05s   | ← Replication: 100x faster indexing

The pre-indexed scenario matches the paper's methodology and validates
the claim of equivalent performance.""" ;
    <https://w3id.org/sciencelive/o/terms/hasLimitationsDescription> """- Vector benchmark tested up to 100 layers (paper used 500)
- Raster pre-indexed scenario simulates but doesn't exactly replicate
  Apache Parquet + Polars implementation
- Missing random misalignment (\"jittering\") from original methodology
- Single hardware configuration tested""" ;
    <https://w3id.org/sciencelive/o/terms/hasOutcomeRepository> <https://doi.org/10.5281/zenodo.18904498> ;
    <https://w3id.org/sciencelive/o/terms/hasValidationStatus> <https://w3id.org/sciencelive/o/terms/PartiallySupported> ;
    <https://w3id.org/sciencelive/o/terms/isOutcomeOf> <https://w3id.org/np/RADCSRkRrlaOzRZ-lkPh1dnvthkhqFz55eGr1wtpW03vk> .
}