Generated: {{ generated_at }}
Document: {{ document_name }}
Runs: {{ runs | length }}
Summary
Pipelines Evaluated
{{ runs | length }}
{{ backends | join(", ") }}
Best KYS
{{ "%.3f" | format(runs[0].kys) }}
{{ runs[0].discovery_model }} / {{ runs[0].extraction_model }}
Most Knowledge
{{ max_graph_size }}
entities + relations
Fastest Pipeline
{{ "%.1f" | format(min_total_time) }}s
end-to-end wall time
Avg Knowledge Rate
{{ "%.2f" | format(avg_knowledge_rate) }}
nodes+edges / second
Total Entities (all)
{{ total_entities }}
across all pipelines
Leaderboard
Sorted by Knowledge Yield Score (KYS) — higher is better
{% for r in runs %} {% endfor %}
# Backend Discovery Model Extraction Model KYS ↓ Entities Relations Graph Size Knowledge Rate Total Time Quality Score Speed Score
{{ loop.index }} {{ r.backend }} {{ r.discovery_model }} {{ r.extraction_model }}
{{ "%.3f" | format(r.kys) }}
{{ r.num_entities }} {{ r.num_relations }} {{ r.graph_size }} {{ "%.2f" | format(r.knowledge_rate) }} /s {{ "%.1f" | format(r.total_time) }}s {{ "%.3f" | format(r.quality_norm) }} {{ "%.3f" | format(r.speed_norm) }}
Visual Analysis

Knowledge Yield Score (KYS)

Pipeline Time Breakdown (seconds)

Knowledge Graph Output

Knowledge Rate (graph elements / second)

Pipeline Details
{% for r in runs %} {% set card_rank = loop.index %}
{{ r.discovery_model }}
extraction: {{ r.extraction_model }}
{{ r.backend }}
#{{ card_rank }}
{{ "%.3f" | format(r.kys) }}
Knowledge Yield Score
Entities
{{ r.num_entities }}
Relations
{{ r.num_relations }}
Total Time
{{ "%.1f" | format(r.total_time) }}s
Knowledge Rate
{{ "%.2f" | format(r.knowledge_rate) }}/s
{% set t = r.total_time %} {% for phase, val, cls in [ ("Schema", r.schema_time, "schema-fill"), ("Rephrase", r.rephrase_time, "rephrase-fill"), ("Extraction", r.extraction_time, "extract-fill"), ("Consolidation", r.consolidation_time, "consolidate-fill") ] %}
{{ phase }}
{{ "%.1f" | format(val) }}s
{% endfor %}
{% endfor %}
Methodology – Knowledge Yield Score (KYS)

The Knowledge Yield Score (KYS) is a composite metric that balances output richness (how much the pipeline extracted) against efficiency (how fast it ran). It is the geometric mean of two normalized sub-scores, analogous to the F₁ score — it penalises runs that excel in only one dimension.

graph_size = num_entities + num_relations
total_time = schema_time + rephrase_time + extraction_time + consolidation_time

quality_norm = graph_size / max(graph_size)  # [0, 1]
speed_norm = min(total_time) / total_time  # [0, 1] — fastest run scores 1.0

KYS = √(quality_norm × speed_norm)  # geometric mean

quality_norm rewards pipelines that produce large, dense knowledge graphs. speed_norm rewards pipelines that finish quickly; the fastest run scores 1.0 and slower runs are penalised proportionally. The geometric mean ensures a pipeline cannot compensate for poor speed with high quality alone — both dimensions must be strong for a high KYS.

Knowledge Rate (graph_size / total_time) is a complementary raw throughput metric expressed in graph elements per second, useful for absolute comparisons independent of the normalization range.