Generated: 2026-04-03 08:35
Document: HumanRights.pdf
Runs: 15
Summary
Pipelines Evaluated
15
anthropic, gemini, ollama, openai
Best KYS
0.751
gemini-3.1-flash-lite-preview / gemini-3.1-flash-lite-preview
Most Knowledge
204
entities + relations
Fastest Pipeline
18.2s
end-to-end wall time
Avg Knowledge Rate
0.96
nodes+edges / second
Total Entities (all)
767
across all pipelines
Leaderboard
Sorted by Knowledge Yield Score (KYS) — higher is better
# Backend Discovery Model Extraction Model KYS ↓ Entities Relations Graph Size Knowledge Rate Total Time Quality Score Speed Score
1 gemini gemini-3.1-flash-lite-preview gemini-3.1-flash-lite-preview
0.751
62 53 115 6.33 /s 18.2s 0.564 1.000
2 anthropic claude-sonnet-4-6 claude-haiku-4-5-20251001
0.395
87 73 160 1.75 /s 91.3s 0.784 0.199
3 openai gpt-4o gpt-4o-mini
0.332
47 52 99 1.24 /s 79.9s 0.485 0.228
4 ollama gemma4:e2b gemma4:e2b
0.245
34 55 89 0.67 /s 132.4s 0.436 0.137
5 ollama mistral:latest mistral:latest
0.244
54 126 180 0.67 /s 268.2s 0.882 0.068
6 ollama gemma3:12b gemma3:4b
0.222
62 45 107 0.55 /s 194.0s 0.524 0.094
7 ollama gemma4:latest gemma4:latest
0.217
85 72 157 0.53 /s 297.4s 0.770 0.061
8 ollama qwen3.5:9b qwen3.5:9b
0.213
99 105 204 0.51 /s 400.7s 1.000 0.045
9 ollama qwen2.5:14b qwen2.5:3b
0.191
37 30 67 0.41 /s 163.3s 0.328 0.111
10 ollama llama3.2:latest llama3.2:latest
0.188
29 36 65 0.40 /s 163.4s 0.319 0.111
11 ollama qwen3:8b qwen3:8b
0.183
58 66 124 0.38 /s 329.4s 0.608 0.055
12 ollama llama3.1:8b llama3.1:8b
0.170
38 46 84 0.33 /s 258.5s 0.412 0.070
13 gemini gemini-3-flash-preview gemini-3-flash-preview
0.163
13 10 23 0.30 /s 76.7s 0.113 0.237
14 ollama glm-4.7-flash:latest glm-4.7-flash:latest
0.156
62 66 128 0.27 /s 466.1s 0.627 0.039
15 gemini gemini-3.1-pro-preview gemini-3.1-pro-preview
0.000
0 0 0 0.00 /s 204.3s 0.000 0.089
Visual Analysis

Knowledge Yield Score (KYS)

Pipeline Time Breakdown (seconds)

Knowledge Graph Output

Knowledge Rate (graph elements / second)

Pipeline Details
gemini-3.1-flash-lite-preview
extraction: gemini-3.1-flash-lite-preview
gemini
#1
0.751
Knowledge Yield Score
Entities
62
Relations
53
Total Time
18.2s
Knowledge Rate
6.33/s
Schema
1.4s
Rephrase
11.7s
Extraction
5.2s
claude-sonnet-4-6
extraction: claude-haiku-4-5-20251001
anthropic
#2
0.395
Knowledge Yield Score
Entities
87
Relations
73
Total Time
91.3s
Knowledge Rate
1.75/s
Schema
7.6s
Rephrase
23.6s
Extraction
60.1s
gpt-4o
extraction: gpt-4o-mini
openai
#3
0.332
Knowledge Yield Score
Entities
47
Relations
52
Total Time
79.9s
Knowledge Rate
1.24/s
Schema
4.8s
Rephrase
48.3s
Extraction
26.8s
gemma4:e2b
extraction: gemma4:e2b
ollama
#4
0.245
Knowledge Yield Score
Entities
34
Relations
55
Total Time
132.4s
Knowledge Rate
0.67/s
Schema
8.5s
Rephrase
30.3s
Extraction
93.6s
mistral:latest
extraction: mistral:latest
ollama
#5
0.244
Knowledge Yield Score
Entities
54
Relations
126
Total Time
268.2s
Knowledge Rate
0.67/s
Schema
0.0s
Rephrase
0.0s
Extraction
268.2s
gemma3:12b
extraction: gemma3:4b
ollama
#6
0.222
Knowledge Yield Score
Entities
62
Relations
45
Total Time
194.0s
Knowledge Rate
0.55/s
Schema
22.4s
Rephrase
36.2s
Extraction
135.3s
gemma4:latest
extraction: gemma4:latest
ollama
#7
0.217
Knowledge Yield Score
Entities
85
Relations
72
Total Time
297.4s
Knowledge Rate
0.53/s
Schema
17.1s
Rephrase
53.7s
Extraction
226.6s
qwen3.5:9b
extraction: qwen3.5:9b
ollama
#8
0.213
Knowledge Yield Score
Entities
99
Relations
105
Total Time
400.7s
Knowledge Rate
0.51/s
Schema
0.0s
Rephrase
0.0s
Extraction
400.7s
qwen2.5:14b
extraction: qwen2.5:3b
ollama
#9
0.191
Knowledge Yield Score
Entities
37
Relations
30
Total Time
163.3s
Knowledge Rate
0.41/s
Schema
22.5s
Rephrase
36.4s
Extraction
104.5s
llama3.2:latest
extraction: llama3.2:latest
ollama
#10
0.188
Knowledge Yield Score
Entities
29
Relations
36
Total Time
163.4s
Knowledge Rate
0.40/s
Schema
8.1s
Rephrase
32.1s
Extraction
123.3s
qwen3:8b
extraction: qwen3:8b
ollama
#11
0.183
Knowledge Yield Score
Entities
58
Relations
66
Total Time
329.4s
Knowledge Rate
0.38/s
Schema
11.9s
Rephrase
61.7s
Extraction
255.8s
llama3.1:8b
extraction: llama3.1:8b
ollama
#12
0.170
Knowledge Yield Score
Entities
38
Relations
46
Total Time
258.5s
Knowledge Rate
0.33/s
Schema
18.5s
Rephrase
39.7s
Extraction
200.3s
gemini-3-flash-preview
extraction: gemini-3-flash-preview
gemini
#13
0.163
Knowledge Yield Score
Entities
13
Relations
10
Total Time
76.7s
Knowledge Rate
0.30/s
Schema
8.3s
Rephrase
47.2s
Extraction
21.2s
glm-4.7-flash:latest
extraction: glm-4.7-flash:latest
ollama
#14
0.156
Knowledge Yield Score
Entities
62
Relations
66
Total Time
466.1s
Knowledge Rate
0.27/s
Schema
22.0s
Rephrase
36.7s
Extraction
407.4s
gemini-3.1-pro-preview
extraction: gemini-3.1-pro-preview
gemini
#15
0.000
Knowledge Yield Score
Entities
0
Relations
0
Total Time
204.3s
Knowledge Rate
0.00/s
Schema
19.4s
Rephrase
145.7s
Extraction
39.2s
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

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.