Ini versi /graph yang benar untuk UNDERCOVER.CO.ID: bukan halaman biasa, tapi live system interface dari knowledge graph engine.

/graph — UNDERCOVER.CO.ID

GRAPH ENGINE INTERFACE v1.0

1. SYSTEM DEFINITION

/graph adalah runtime access layer untuk seluruh knowledge system UNDERCOVER.CO.ID.

Fungsi utamanya:

expose struktur graph (node + edge)
memungkinkan traversal knowledge secara real-time
menjadi jembatan antara /center (kernel rules) dan /entity /topic /query /evidence

Ini bukan sitemap. Ini queryable knowledge infrastructure endpoint.

2. CORE PURPOSE

/graph menjalankan 3 fungsi inti:

2.1 Structural Visibility

Menampilkan bagaimana sistem benar-benar tersusun:

node apa saja yang aktif
bagaimana mereka terhubung
seberapa kuat hubungan antar node

2.2 Query Execution Layer

Mengubah input menjadi traversal graph:

query → intent → node traversal → result set

2.3 System Debug Interface

Untuk audit:

orphan node detection
weak relationship detection
authority imbalance detection

3. DATA MODEL

3.1 NODE OBJECT

{
  "id": "string",
  "type": "entity | topic | query | evidence | index",
  "name": "string",
  "description": "string",
  "authority_score": 0.0,
  "embedding_vector": [],
  "metadata": {
    "created_at": "timestamp",
    "updated_at": "timestamp"
  }
}

3.2 EDGE OBJECT

{
  "from": "node_id",
  "to": "node_id",
  "relation": "BELONGS_TO | CONTAINS | TARGETS | RESOLVES_TO | VALIDATES | SUPPORTS | RELATED_TO",
  "strength": 0.0,
  "context": "string"
}

4. GRAPH OPERATIONS

4.1 Get Full Graph (System View)

GET /graph

Returns:

all nodes (filtered by system rules)
all edges
global connectivity map

4.2 Get Node Subgraph

GET /graph/node/{id}

Returns:

target node
1-hop connections
2-hop expanded relationships
authority ranking context

4.3 Query Graph (Core Function)

POST /graph/resolve

Request:

{
  "query": "string"
}

Response:

{
  "intent": "parsed_intent",
  "matched_topics": [],
  "resolved_entities": [],
  "ranked_output": []
}

4.4 Relationship Explorer

GET /graph/explore?node_id={id}

Returns:

full relationship tree
edge strength map
cluster affiliation

4.5 Authority Map

GET /graph/authority

Returns:

top central nodes
weak nodes
orphan nodes
cluster density score

5. TRAVERSAL ENGINE

Core Logic Flow

QUERY INPUT
   ↓
Intent Parsing
   ↓
Seed Node Selection
   ↓
Graph Expansion (Depth 1–3)
   ↓
Edge Weight Scoring
   ↓
Authority Computation
   ↓
Evidence Filtering
   ↓
Ranked Output

Depth Rules

Depth 1 → direct entity match
Depth 2 → topic expansion
Depth 3 → cross-topic linkage
Depth 4+ → suppressed (noise control)

6. INTEGRATION WITH /CENTER

/graph does NOT define rules.

It only executes rules from /center.

Relationship:

/center = system brain (logic + policy)
/graph = execution engine (runtime system)

If /center changes:
→ /graph behavior changes automatically

7. AUTHORITY CALCULATION MODEL

authority(node) =
    centrality_score +
    inbound_edges_weight +
    cross_cluster_connections +
    evidence_support_density +
    query_activation_frequency

8. SYSTEM BEHAVIOR RULES

Rule 1 — No Isolated Nodes

Every node must have at least 1 edge.

Rule 2 — Entity Priority

Entity nodes dominate ranking output.

Rule 3 — Evidence is Filter Only

Evidence cannot override entity authority.

Rule 4 — Query Always Resolves

Every query must resolve into at least 1 entity node.

9. OUTPUT MODES

/graph supports 3 views:

9.1 Raw Graph Mode

Full node-edge structure

9.2 Ranked Mode

Sorted by authority score

9.3 Cluster Mode

Grouped by topic clusters

10. SYSTEM ROLE SUMMARY

/graph is:

not content
not index
not navigation page

It is:

LIVE EXECUTION LAYER OF UNDERCOVER KNOWLEDGE SYSTEM

FINAL ARCHITECTURE POSITION

/center → system rules + brain
/graph → execution engine + runtime
/entity → atomic knowledge
/topic → clustering layer
/query → intent layer
/evidence → validation layer
/index → aggregation layer

STRUCTURED SUMMARY

The /graph page functions as the runtime execution interface for UNDERCOVER.CO.ID’s knowledge system, enabling real-time traversal of entities, topics, queries, and evidence through a structured node-edge architecture. It operates under rules defined by /center and provides query resolution, relationship mapping, and authority scoring across the system, making it the core execution layer of the AI-first knowledge graph infrastructure.