Files
llm-wiki/tests/expected/graph-data-sample.json
2026-07-12 21:26:08 +08:00

397 lines
9.4 KiB
JSON

{
"meta": {
"build_date": "2026-01-01T00:00:00Z",
"wiki_title": "测试知识库",
"total_nodes": 8,
"total_edges": 13,
"initial_view": [
"Attention",
"Transformer",
"GPT",
"arch",
"Decoder",
"Encoder",
"finetune",
"paper"
],
"degraded": false,
"insights_degraded": false
},
"nodes": [
{
"id": "Attention",
"label": "Attention",
"type": "entity",
"community": "Attention",
"content": "# Attention\n\n注意力机制是 Transformer 的核心。参见 [[Transformer]]。"
},
{
"id": "Decoder",
"label": "Decoder",
"type": "entity",
"community": "arch",
"content": "# Decoder\n\n解码器组件。与 [[Transformer]] <!-- confidence: INFERRED --> 的解码端对应。"
},
{
"id": "Encoder",
"label": "Encoder",
"type": "entity",
"community": "arch",
"content": "# Encoder\n\n编码器组件。属于 [[Transformer]] 架构的一部分。"
},
{
"id": "GPT",
"label": "GPT",
"type": "entity",
"community": "finetune",
"content": "# GPT\n\nGPT 系列基于 [[Transformer]] <!-- confidence: AMBIGUOUS --> 的解码器架构。"
},
{
"id": "Transformer",
"label": "Transformer",
"type": "entity",
"community": "arch",
"content": "# Transformer\n\nTransformer 是一种基于自注意力机制的序列到序列模型架构。\n\n## 核心组件\n\n- [[Attention]] — 自注意力机制\n- [[Encoder]] — 编码器\n- [[Decoder]] — 解码器"
},
{
"id": "arch",
"label": "深度学习架构",
"type": "topic",
"community": "arch",
"content": "# 深度学习架构\n\n本主题涵盖深度学习的核心架构组件。\n\n- [[Transformer]]\n- [[Attention]]\n- [[Encoder]]\n- [[Decoder]]"
},
{
"id": "finetune",
"label": "微调技术",
"type": "topic",
"community": "finetune",
"content": "# 微调技术\n\n本主题涵盖模型微调相关技术。\n\n- [[GPT]]"
},
{
"id": "paper",
"label": "Attention Is All You Need",
"type": "source",
"community": "Attention",
"content": "# Attention Is All You Need\n\n原始论文提出了自注意力机制。\n\n- [[Attention]] <!-- confidence: EXTRACTED -->"
}
],
"edges": [
{
"id": "e1",
"from": "Attention",
"to": "Transformer",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.733,
"source_signal_available": true,
"signals": {
"co_citation": 0.2,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e2",
"from": "Decoder",
"to": "Transformer",
"type": "INFERRED",
"confidence": "INFERRED",
"relation_type": "依赖",
"weight": 0.733,
"source_signal_available": true,
"signals": {
"co_citation": 0.2,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e3",
"from": "Encoder",
"to": "Transformer",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.733,
"source_signal_available": true,
"signals": {
"co_citation": 0.2,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e4",
"from": "GPT",
"to": "Transformer",
"type": "AMBIGUOUS",
"confidence": "AMBIGUOUS",
"relation_type": "依赖",
"weight": 0.333,
"source_signal_available": true,
"signals": {
"co_citation": 0,
"source_overlap": 0,
"type_affinity": 1
}
},
{
"id": "e5",
"from": "Transformer",
"to": "Attention",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.733,
"source_signal_available": true,
"signals": {
"co_citation": 0.2,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e6",
"from": "Transformer",
"to": "Decoder",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.733,
"source_signal_available": true,
"signals": {
"co_citation": 0.2,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e7",
"from": "Transformer",
"to": "Encoder",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.733,
"source_signal_available": true,
"signals": {
"co_citation": 0.2,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e8",
"from": "arch",
"to": "Attention",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.667,
"source_signal_available": true,
"signals": {
"co_citation": 0,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e9",
"from": "arch",
"to": "Decoder",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.667,
"source_signal_available": true,
"signals": {
"co_citation": 0,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e10",
"from": "arch",
"to": "Encoder",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.667,
"source_signal_available": true,
"signals": {
"co_citation": 0,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e11",
"from": "arch",
"to": "Transformer",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.667,
"source_signal_available": true,
"signals": {
"co_citation": 0,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e12",
"from": "finetune",
"to": "GPT",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.667,
"source_signal_available": true,
"signals": {
"co_citation": 0,
"source_overlap": 1,
"type_affinity": 1
}
},
{
"id": "e13",
"from": "paper",
"to": "Attention",
"type": "EXTRACTED",
"confidence": "EXTRACTED",
"relation_type": "依赖",
"weight": 0.533,
"source_signal_available": true,
"signals": {
"co_citation": 0,
"source_overlap": 1,
"type_affinity": 0.6
}
}
],
"insights": {
"surprising_connections": [],
"isolated_nodes": [
{
"id": "finetune",
"label": "微调技术",
"degree": 1,
"community": "finetune"
},
{
"id": "paper",
"label": "Attention Is All You Need",
"degree": 1,
"community": "Attention"
}
],
"bridge_nodes": [
{
"id": "Transformer",
"label": "Transformer",
"community": "arch",
"connected_communities": [
"Attention",
"finetune"
],
"community_count": 2
}
],
"sparse_communities": [],
"meta": {
"degraded": false,
"node_count": 8,
"edge_count": 13,
"max_insight_nodes": 250,
"max_insight_edges": 1000
}
},
"learning": {
"version": 1,
"entry": {
"recommended_start_node_id": "Transformer",
"recommended_start_reason": "community_hub",
"default_mode": "global"
},
"views": {
"path": {
"enabled": true,
"start_node_id": "Transformer",
"node_ids": [
"Transformer",
"Decoder",
"Encoder",
"arch"
],
"degraded": false
},
"community": {
"enabled": true,
"community_id": "arch",
"label": "深度学习架构",
"node_ids": [
"Decoder",
"Encoder",
"Transformer",
"arch"
],
"is_weak": false,
"degraded": false
},
"global": {
"enabled": true,
"node_ids": [
"Transformer",
"arch",
"Attention",
"Decoder",
"Encoder",
"GPT",
"finetune",
"paper"
],
"degraded": false
}
},
"communities": [
{
"id": "arch",
"label": "深度学习架构",
"node_count": 4,
"source_count": 0,
"internal_edge_weight": 4.933,
"is_primary": true,
"is_weak": false,
"recommended_start_node_id": "Transformer"
},
{
"id": "finetune",
"label": "微调技术",
"node_count": 2,
"source_count": 0,
"internal_edge_weight": 0.667,
"is_primary": false,
"is_weak": true,
"recommended_start_node_id": "GPT"
},
{
"id": "Attention",
"label": "Attention",
"node_count": 2,
"source_count": 1,
"internal_edge_weight": 0.533,
"is_primary": false,
"is_weak": true,
"recommended_start_node_id": "Attention"
}
],
"degraded": {
"path_to_community": false,
"community_to_global": false
}
}
}