PAPER_ARTIFACT_MAP.md

Paper Artifact Map

This document maps each paper claim to the benchmark evidence, output files, and intended paper destination.

Historical Materialized Subquery Claim

Paper claim	Benchmark	Output files	Figure/table destination	Notes and caveats
Historical materialized subqueries lower to the historical materialization path and behave similarly to explicit DEFINE BASELINE	Historical Materialized Subquery Benchmark	target/paper_benchmarks/historical_materialized_subquery/20260619_110106/historical_materialized_subquery.md	Appendix table for explicit vs nested historical subquery comparison	baseline_bindings is a sanity check, not a claim metric. first_result_latency_ms_avg is n/a in this benchmark.

Unified Hybrid Execution Claim

Paper claim	Benchmark	Output files	Figure/table destination	Notes and caveats
Unified hybrid execution reduces coordination complexity and avoids external transfer relative to the decomposed baseline	Hybrid Coordination Benchmark	target/paper_benchmarks/paper_hybrid_coordination/20260619_110106/paper_hybrid_coordination.summary.csv, target/paper_benchmarks/paper_hybrid_coordination/20260619_110106/paper_hybrid_coordination.raw.jsonl	Figure 2: unified vs decomposed hybrid execution latency; Figure 3: external transfer; main coordination table	The decomposed row label in raw output is implementation-specific. historical_equivalence_rate and hybrid_equivalence_rate for the decomposed row are schema/design zeros and should not be interpreted as failures.

Historical Scaling Claim

Paper claim	Benchmark	Output files	Figure/table destination	Notes and caveats
Historical query latency changes with dataset size, with point lookups and fixed windows remaining nearly flat while scan-heavy queries grow strongly	Historical Scaling Benchmark	target/paper_benchmarks/paper_historical_scaling_full/20260619_110323/paper_historical_scaling.summary.csv, target/paper_benchmarks/paper_historical_scaling_full/20260619_110323/paper_historical_scaling.raw.jsonl, target/paper_benchmarks/paper_historical_scaling_full/20260619_110323/paper_historical_scaling.fit.csv	Figure 1: historical query latency by dataset size; main/appendix historical scaling table	hybrid_baseline_lookup is the noisiest query at 5M quads. The fit CSV is supporting evidence, not the primary result.

Output Package Notes

• Keep the figure outputs and compact tables for the submission package.
• Keep raw CSV, JSONL, fit CSV, and generated citybench_*.nq datasets local only.
• Use the benchmark summaries and tables for review; do not cite raw artifacts directly in the paper unless needed for appendix verification.