Phase 6: Insight Analysis¶

Module: src/insight_analyzer.py
Estimated Time: ~1 day

Objective¶

Analyze the knowledge graph using graph algorithms to identify cross-disciplinary patterns, hub concepts, field influence dynamics, and generate a structured Markdown report.

Analysis Components¶

1. Hub Concept Analysis¶

Identifies bridge concepts that connect the most different scientific fields:

Counts the number of distinct fields each concept connects to
Ranks concepts by cross-field connectivity
Top concepts are listed as "interdisciplinary hubs"

Example result:

Concept	Connected Fields
Protein Degradation Mechanisms	Biology, Chemistry, Physics
Ubiquitin-Proteasome Pathway	Biology, Chemistry
Fixed Cost Analysis	Economics, Business

2. Field Influence Analysis¶

Measures which fields export concepts to others and which import:

Counts outgoing CROSS_INSPIRED edges per source field
Counts incoming CROSS_INSPIRED edges per target field
Identifies net exporters and importers of concepts

3. Temporal Pattern Analysis¶

Extracts the timeline of cross-disciplinary migration events:

Lists all CROSS_INSPIRED edges with timestamps
Identifies acceleration trends in cross-field inspiration
Tracks which decades had the most cross-pollination

4. Key Pathway Analysis¶

Finds shortest paths between concepts that cross field boundaries:

Uses NetworkX shortest path algorithms
Filters for paths of length 3–6 hops
Requires at least 2 different fields along the path

Report Structure¶

The generated report (output/reports/insight_report.md) contains 6 sections:

Graph Overview — Node/edge counts, node type breakdown
Top Hub Concepts — Ranked list of cross-disciplinary bridge concepts
Field Influence — Import/export analysis by field
Cross-Discipline Timeline — Temporal listing of migration events
Key Pathways — Notable cross-field concept paths
Methodology — How the analysis was performed

Output Files¶

File	Format	Description
`insight_report.md`	Markdown	Human-readable analysis report
`insight_report.json`	JSON	Machine-readable analysis data

JSON Structure¶

{
  "graph_overview": {
    "total_nodes": 97,
    "total_edges": 181,
    "node_types": {"Laureate": 5, "Work": 25, "Concept": 51, ...},
    "edge_types": {"WON_AWARD": 5, "AUTHORED": 25, ...}
  },
  "hub_concepts": [...],
  "field_influence": {...},
  "temporal_patterns": [...],
  "key_pathways": [...]
}

Running¶

# Via pipeline
uv run python main.py --phase 6

# Standalone
uv run python -m src.insight_analyzer