Finding the Hubs¶

Part 3: Exploring Your Knowledge Graph | Chapter 5 of 6

Prerequisites: Chapters 3.1 through 3.4 Time: ~20 minutes

Quick Recap¶

Last chapter, you used WHERE NOT EXISTS and existential subqueries to find the orphans: 8 concepts with zero connections, sitting in the graph doing nothing. You also learned to count each concept's connections with COUNT { MATCH ... }, which gave you a rough sense of who's busy and who's lonely.

That counting trick brought up an obvious question. Let's answer it.

Pop Quiz¶

What's the most connected concept in your Chapter 5 graph?

You'd guess Encapsulation, right? It's literally the chapter title. The whole chapter is organized around it. Every other concept exists in service of Encapsulation.

Let's find out.

Counting Every Concept's Connections¶

Start simple. For each concept, count how many relationships touch it (in any direction), and sort by that count:

MATCH (c:Concept)-[r]-()
RETURN c.name, count(r) AS connections
ORDER BY connections DESC
LIMIT 10

Result:

Read that table again. Slowly.

Properties: 23 connections.

Encapsulation: 8.

That's not close. Properties has nearly three times the connections of the chapter's own title concept.

The Concept Everything Connects To¶

Before we figure out why, let's confirm. Pull just the top result:

MATCH (c:Concept)-[r]-()
RETURN c.name, count(r) AS connections
ORDER BY connections DESC
LIMIT 1

Result:

Properties. Not even close.

Here's what's happening. Properties is the concrete mechanism that every other concept in the chapter talks about, depends on, refines, or challenges. Getters use Properties. Setters modify Properties. Validation guards Properties. Access Control restricts Properties. Encapsulation is the goal, but Properties is the crossroads where all those concepts actually meet.

The concept everything connects to isn't the abstract idea at the top. It's the implementation detail in the middle.

Properties sits at the center. Everything radiates out from it, or points into it. That makes it a hub: the concept everything connects to.

Direction Matters: In vs Out¶

Not all connections are the same. In a directed graph, an arrow pointing in means something different from an arrow pointing out.

A concept with lots of incoming connections is something other concepts depend on. A concept with lots of outgoing connections is something that reaches out to affect other things.

Let's split the count:

MATCH (c:Concept)
OPTIONAL MATCH (c)-[out]->()
OPTIONAL MATCH (c)<-[in]-()
RETURN c.name,
       count(DISTINCT out) AS out_connections,
       count(DISTINCT in) AS in_connections,
       count(DISTINCT out) + count(DISTINCT in) AS total
ORDER BY total DESC
LIMIT 10

Result:

Two things jump out.

First: Properties has 15 incoming connections and only 8 outgoing. That means 15 other concepts point TO Properties. It's a dependency magnet. Things need it.

Second: look at Dangerous Setters. 5 outgoing, 1 incoming. It's almost entirely an emitter: it reaches out to affect other things (enables Invalid State, challenges Encapsulation), but almost nothing depends on it. Dangerous Setters is a cause, not a dependency.

Why This Matters¶

Here's the thing that makes this more than a Cypher exercise.

If someone asks your system about "data hiding" or "encapsulation," a naive text search would match the Encapsulation concept. That's a reasonable match. But from a graph structure perspective, Properties is far more important. It sits at the junction of 23 relationships. It connects to getters, setters, validation, access control, and encapsulation itself.

If you want to understand how concepts in this chapter relate to each other, Properties is the better starting point. Not because it's a fancier concept, but because it's the crossroads. More paths run through Properties than through any other concept.

The abstract goal isn't the structural center. The concrete mechanism is. This is a pattern you'll see in other chapters too: the chapter title concept organizes the narrative, but some implementation detail underneath it is the real connective tissue.

The Words You Just Earned¶

You've been counting connections and finding the concept everything connects to. Let's name what you've been doing.

Degree: the number of connections a concept has. Properties has degree 23. You can split it into in-degree (incoming connections, 15 for Properties) and out-degree (outgoing connections, 8 for Properties).

Hub: a concept with unusually high degree. It's the node in the graph where lots of paths converge. Properties is this graph's hub.

Centrality: the formal name for "importance measured by connections." When someone says "degree centrality," they mean exactly what you just computed: rank the nodes by their number of connections, and the top ones are the most central. There are fancier kinds of centrality (betweenness, closeness, eigenvector), but degree centrality is the simplest and the one you earned today.

Exercises¶

Exercise 1: The In-Degree Champion¶

Problem: Find the concept with the highest in-degree (the most things pointing TO it). Is it the same concept as the one with the highest total degree?

Hint: You already have the query for this. Just change the ORDER BY.

MATCH (c:Concept)
OPTIONAL MATCH (c)<-[in]-()
RETURN c.name, count(DISTINCT in) AS in_degree
ORDER BY in_degree DESC
LIMIT 5

Exercise 2: The Emitters¶

Problem: Find all concepts where out-degree is strictly greater than in-degree. What do these concepts tend to be?

Hint: Use WITH to compute both counts, then filter with WHERE.

MATCH (c:Concept)
OPTIONAL MATCH (c)-[out]->()
OPTIONAL MATCH (c)<-[in]-()
WITH c.name AS name,
     count(DISTINCT out) AS out_deg,
     count(DISTINCT in) AS in_deg
WHERE out_deg > in_deg
RETURN name, out_deg, in_deg
ORDER BY out_deg DESC

Exercise 3: The Structural Backbone¶

Problem: Find all concepts with total degree greater than 5. These are the concepts holding your graph together, its structural backbone.

Hint: This is the first query from the chapter, with a filter added.

MATCH (c:Concept)-[r]-()
WITH c.name AS name, count(r) AS degree
WHERE degree > 5
RETURN name, degree
ORDER BY degree DESC

What You Learned¶

• Counting connections reveals which concepts are structurally important, and the answer isn't always what you'd expect
• In-degree vs out-degree tells you whether a concept is a dependency (things point to it) or a cause (it points to things)
• The hub isn't the abstract goal, it's the concrete mechanism: Properties outranks Encapsulation by nearly 3x because it's the implementation crossroads
• Degree, hub, centrality: three terms you earned by doing the work first, then naming it

Next Up¶

You found the hub. You know Properties is the crossroads. But here's a harder question: what are ALL the paths from Dangerous Setters to Production Bug? There might be more than one. They might be different lengths. And when two paths are the same length, which one matters most?

That last question is the one Cypher can't answer. Let's find out why.

Chapter 3.6: Paths and Distances