Research-Stack/scripts/analyze_concept_graph.py

#!/usr/bin/env -S uv run
# /// script
# requires-python = ">=3.11"
# dependencies = ["gremlinpython", "python-dotenv"]
# ///
"""
analyze_concept_graph.py — Graph topology analysis for cornfield concepts.

Queries:
  Q1: Degree distribution (out + in) for all cornfield concepts
  Q2: Bridge concepts (relates_to degree > 2)
  Q3: Missing depends_on edges (has lean module but no edge)
  Q4: Active Lean modules most referenced by cornfield concepts
  Q5: relates_to pairs among cornfield concepts (cluster analysis)
"""
import os
import sys
from pathlib import Path
from dotenv import load_dotenv

load_dotenv(Path(__file__).parent.parent / ".env.gremlin")

from gremlin_python.driver import client as gc, serializer

c = gc.Client(
    os.environ["GREMLIN_ENDPOINT"], "g",
    username=os.environ["GREMLIN_USERNAME"],
    password=os.environ["GREMLIN_PASSWORD"],
    message_serializer=serializer.GraphSONSerializersV2d0(),
)

def q(query, bindings=None):
    try:
        result = c.submitAsync(query, bindings or {}).result().all().result()
        return result
    except Exception as e:
        print(f"  [ERROR] {e}")
        return []

SEP = "═" * 68

print(SEP)
print("Q1: CORNFIELD CONCEPT DEGREE DISTRIBUTION")
print("─" * 68)

r = q(
    "g.V().has('cohort','cornfield')"
    ".project('vid','out_deg','in_deg')"
    ".by(id())"
    ".by(outE().count())"
    ".by(inE().count())"
)

if r:
    # Sort by total degree desc
    rows = sorted(r, key=lambda x: x['out_deg'] + x['in_deg'])
    print(f"  {'ID':<40} {'out':>4} {'in':>4} {'total':>6}")
    print("  " + "-" * 60)
    isolated = []
    for row in rows:
        vid = str(row['vid'])
        out_d = row['out_deg']
        in_d = row['in_deg']
        total = out_d + in_d
        flag = " *** ISOLATED" if total <= 1 else (" * LOW" if total <= 2 else "")
        print(f"  {vid:<40} {out_d:>4} {in_d:>4} {total:>6}{flag}")
        if total <= 1:
            isolated.append(vid)
    print()
    print(f"  Total cornfield concepts queried: {len(rows)}")
    print(f"  Isolated (degree <= 1): {isolated}")
else:
    print("  No results or error.")

print()
print(SEP)
print("Q2: BRIDGE CONCEPTS (relates_to out-degree > 2)")
print("─" * 68)

r2 = q(
    "g.V().has('cohort','cornfield')"
    ".filter(outE('relates_to').count().is(gt(2)))"
    ".project('vid','rt_count')"
    ".by(id())"
    ".by(outE('relates_to').count())"
    ".order().by(select('rt_count'), decr)"
)

if r2:
    for row in r2:
        print(f"  {str(row['vid']):<40}  relates_to_out={row['rt_count']}")
else:
    print("  None found (no concept has > 2 outgoing relates_to edges).")

print()
print(SEP)
print("Q3: MISSING depends_on EDGES (has lean module, no depends_on edge)")
print("─" * 68)

r3 = q(
    "g.V().has('cohort','cornfield')"
    ".has('lean_status', within('LIBRARY','ARCHIVED','PARTIALLY_ACTIVE','PROVEN','DEFINED'))"
    ".not(outE('depends_on'))"
    ".project('vid','lean_status')"
    ".by(id())"
    ".by('lean_status')"
)

if r3:
    for row in r3:
        print(f"  {str(row['vid']):<40}  lean_status={row.get('lean_status','?')}")
    print(f"\n  Count: {len(r3)} concepts with Lean modules but no depends_on edge")
else:
    print("  All concepts with Lean modules already have depends_on edges, or none found.")

print()
print(SEP)
print("Q4: TOP ACTIVE LEAN MODULES (most cornfield depends_on edges incoming)")
print("─" * 68)

r4 = q(
    "g.V().has('cohort','cornfield').out('depends_on')"
    ".hasLabel('module')"
    ".groupCount().by(id())"
    ".order(local).by(values, decr).limit(local, 15)"
)

if r4 and len(r4) > 0:
    result_map = r4[0] if isinstance(r4[0], dict) else {}
    for mod_id, count in sorted(result_map.items(), key=lambda x: -x[1]):
        print(f"  {str(mod_id):<55}  count={count}")
else:
    print("  No depends_on edges to modules found, or error.")

print()
print(SEP)
print("Q5: CORNFIELD CONCEPT CLUSTER ANALYSIS (relates_to pairs)")
print("─" * 68)

r5 = q(
    "g.V().has('cohort','cornfield').as('a')"
    ".out('relates_to').has('cohort','cornfield').as('b')"
    ".select('a','b').by(id())"
)

if r5:
    print(f"  Total relates_to edges between cornfield concepts: {len(r5)}")
    print()
    # Build adjacency for cluster analysis
    adj = {}
    for row in r5:
        a = str(row['a'])
        b = str(row['b'])
        adj.setdefault(a, set()).add(b)
        adj.setdefault(b, set()).add(a)

    print("  Pairs found:")
    for row in r5:
        a = str(row['a'])
        b = str(row['b'])
        print(f"    {a:<40} → {b}")

    print()
    # Find concepts with multiple relates_to neighbors
    print("  Cluster hubs (2+ cornfield neighbours):")
    for node, neighbors in sorted(adj.items(), key=lambda x: -len(x[1])):
        if len(neighbors) >= 2:
            print(f"    {node:<40} ({len(neighbors)} neighbors): {sorted(neighbors)[:5]}")
else:
    print("  No relates_to edges between cornfield concepts found in graph.")

print()
print(SEP)
print("Q6: ALL EDGE LABELS ON CORNFIELD VERTICES (to see what edges exist)")
print("─" * 68)

r6 = q(
    "g.V().has('cohort','cornfield').bothE()"
    ".groupCount().by(label())"
)

if r6 and r6[0]:
    for label, count in sorted(r6[0].items()):
        print(f"  {label:<30} count={count}")
else:
    print("  No edges found on cornfield concepts.")

print()
print(SEP)
print("Q7: FULL DEGREE TABLE SORTED BY TOTAL (ascending — find lowest)")
print("─" * 68)

# Show all vertex IDs with their property values to match to concept names
r7 = q(
    "g.V().has('cohort','cornfield')"
    ".project('vid','name','lean_status','tags')"
    ".by(id())"
    ".by(coalesce(values('name'), constant('?')))"
    ".by(coalesce(values('lean_status'), constant('?')))"
    ".by(coalesce(values('tags'), constant('?')))"
)

if r7:
    print(f"  Total concepts: {len(r7)}")
    for row in r7:
        print(f"  id={str(row['vid']):<45} name={str(row.get('name','?')):<40} status={str(row.get('lean_status','?')):<25} tags={str(row.get('tags','?'))[:60]}")
else:
    print("  No property results.")

c.close()
print()
print("Done.")