Taxonomy Distribution Report

Overview

• Total Concepts: 300
• Number of Taxonomies: 15
• Average Concepts per Taxonomy: 20.0

Distribution Summary

Category	TaxonomyID	Count	Percentage	Status
GNN	GNN	44	14.7%	✅
Foundation Concepts - Prerequisites	FOUND	37	12.3%	✅
ALGO	ALGO	33	11.0%	✅
KG	KG	25	8.3%	✅
APP	APP	23	7.7%	✅
PREREQ	PREREQ	20	6.7%	✅
TRAIN	TRAIN	20	6.7%	✅
Advanced Topics	ADV	18	6.0%	✅
EMB	EMB	17	5.7%	✅
THEORY	THEORY	17	5.7%	✅
TRANS	TRANS	12	4.0%	✅
SCALE	SCALE	10	3.3%	✅
TOOLS	TOOLS	9	3.0%	✅
HETERO	HETERO	8	2.7%	ℹ️ Under
GEN	GEN	7	2.3%	ℹ️ Under

Visual Distribution

GNN                       ███████  44 ( 14.7%)
Foundation Concepts - Pre ██████  37 ( 12.3%)
ALGO                      █████  33 ( 11.0%)
KG                        ████  25 (  8.3%)
APP                       ███  23 (  7.7%)
PREREQ                    ███  20 (  6.7%)
TRAIN                     ███  20 (  6.7%)
Advanced Topics           ███  18 (  6.0%)
EMB                       ██  17 (  5.7%)
THEORY                    ██  17 (  5.7%)
TRANS                     ██  12 (  4.0%)
SCALE                     █  10 (  3.3%)
TOOLS                     █   9 (  3.0%)
HETERO                    █   8 (  2.7%)
GEN                       █   7 (  2.3%)

Balance Analysis

✅ No Over-Represented Categories

All categories are under the 30% threshold. Good balance!

ℹ️ Under-Represented Categories (<3%)

• HETERO (HETERO): 8 concepts (2.7%)
• Note: Small categories are acceptable for specialized topics
• GEN (GEN): 7 concepts (2.3%)
• Note: Small categories are acceptable for specialized topics

Category Details

GNN (GNN)

Count: 44 concepts (14.7%)

Concepts:

• 1. Graph Neural Network (GNN)
• 1. Message Passing Neural Net
• 1. Message Function
• 1. Aggregation Function
• 1. Update Function
• 1. Graph Convolutional Network
• 1. GraphSAGE
• 1. Graph Attention Network (GAT)
• 1. Attention Mechanism (Graph)
• 1. Multi-Head Attention (Graph)
• 1. Graph Isomorphism Network
• 1. Sum Aggregation
• 1. Mean Aggregation
• 1. Max Aggregation
• 1. Neighborhood Aggregation
• ...and 29 more

Foundation Concepts - Prerequisites (FOUND)

Count: 37 concepts (12.3%)

Concepts:

• 1. Graph (Undirected)
• 1. Graph (Directed)
• 1. Node (Vertex)
• 1. Edge (Link)
• 1. Adjacency Matrix
• 1. Node Degree
• 1. In-Degree
• 1. Out-Degree
• 1. Degree Distribution
• 1. Power-Law Degree Distribution
• 1. Graph Path
• 1. Shortest Path
• 1. Graph Diameter
• 1. Connected Component
• 1. Strongly Connected Component
• ...and 22 more

ALGO (ALGO)

Count: 33 concepts (11.0%)

Concepts:

• 1. Local Clustering Coefficient
• 1. Betweenness Centrality
• 1. Closeness Centrality
• 1. Eigenvector Centrality
• 1. PageRank
• 1. Personalized PageRank
• 1. HITS Algorithm
• 1. Random Walk
• 1. Stationary Distribution
• 1. Power Iteration
• 1. Teleportation (PageRank)
• 1. Community Detection
• 1. Modularity (Network)
• 1. Louvain Algorithm
• 1. Girvan-Newman Algorithm
• ...and 18 more

KG (KG)

Count: 25 concepts (8.3%)

Concepts:

• 1. Knowledge Graph
• 1. KG Entity
• 1. KG Relation
• 1. KG Triple
• 1. KG Completion
• 1. Link Prediction (KG)
• 1. TransE
• 1. TransR
• 1. DistMult
• 1. ComplEx
• 1. RotatE
• 1. KG Embedding Geometry
• 1. Bilinear KG Model
• 1. Symmetry (Relation Pattern)
• 1. Antisymmetry (Relation)
• ...and 10 more

APP (APP)

Count: 23 concepts (7.7%)

Concepts:

• 1. Recommender System (Graph)
• 1. Collaborative Filtering
• 1. Matrix Factorization (Rec)
• 1. Neural Collaborative Filter
• 1. LightGCN
• 1. PinSage
• 1. Drug-Drug Interaction
• 1. Protein-Protein Interaction
• 1. Molecular Graph
• 1. Drug Discovery with GNNs
• 1. Social Network Analysis
• 1. Fraud Detection (Graph)
• 1. Traffic Forecasting (Graph)
• 1. Scene Graph
• 1. Relational Database as Graph
• ...and 8 more

PREREQ (PREREQ)

Count: 20 concepts (6.7%)

Concepts:

• 1. Matrix Multiplication
• 1. Matrix Transpose
• 1. Eigenvalue Decomposition
• 1. Eigenvector
• 1. Symmetric Matrix
• 1. Positive Semi-Definite Matrix
• 1. Singular Value Decomp (SVD)
• 1. Matrix Rank
• 1. Dot Product
• 1. Cosine Similarity
• 1. Gradient Descent
• 1. Backpropagation
• 1. Chain Rule (Calculus)
• 1. Automatic Differentiation
• 1. PyTorch Tensor
• ...and 5 more

TRAIN (TRAIN)

Count: 20 concepts (6.7%)

Concepts:

• 1. Cross-Entropy (Node Class)
• 1. Binary Cross-Entropy (LP)
• 1. Negative Sampling (LP)
• 1. Data Augmentation (Graph)
• 1. Dropout (GNN)
• 1. Batch Normalization (GNN)
• 1. Layer Normalization (GNN)
• 1. DropEdge
• 1. PairNorm
• 1. Early Stopping (GNN)
• 1. Train/Val/Test Split (Trans)
• 1. Train/Val/Test Split (Ind)
• 1. Curriculum Learning (Graph)
• 1. Contrastive Loss
• 1. Graph Augmentation (SSL)
• ...and 5 more

Advanced Topics (ADV)

Count: 18 concepts (6.0%)

Concepts:

• 1. In-Context Learning (Graphs)
• 1. Conformalized GNN
• 1. Graph Foundation Model
• 1. LLM + GNN Integration
• 1. Text-Attributed Graph
• 1. Graph Instruction Tuning
• 1. One-For-All (OFA) Model
• 1. Agent Memory as Graph
• 1. Tool-Use Graph
• 1. Relational Deep Learning
• 1. RelBench
• 1. RelGNN
• 1. Self-Supervised Learning
• 1. Deep Graph Infomax (DGI)
• 1. Graph Contrastive Learning
• ...and 3 more

EMB (EMB)

Count: 17 concepts (5.7%)

Concepts:

• 1. Node Embedding
• 1. Embedding Space
• 1. Encoder-Decoder Framework
• 1. Shallow Embedding
• 1. Matrix Factorization (Graph)
• 1. DeepWalk
• 1. node2vec
• 1. Biased Random Walk
• 1. BFS Strategy (node2vec)
• 1. DFS Strategy (node2vec)
• 1. Skip-Gram Model
• 1. Negative Sampling
• 1. LINE Embedding
• 1. Transductive Learning
• 1. Inductive Learning
• ...and 2 more

THEORY (THEORY)

Count: 17 concepts (5.7%)

Concepts:

• 1. Weisfeiler-Lehman (WL) Test
• 1. 1-WL Test
• 1. k-WL Test
• 1. GNN Expressiveness
• 1. Graph Isomorphism Problem
• 1. GNN Distinguishing Power
• 1. Over-Smoothing
• 1. Over-Squashing
• 1. GNN Bottleneck
• 1. Graph Equivariance
• 1. Graph Invariance
• 1. Permutation Invariance
• 1. Permutation Equivariance
• 1. Position-Aware GNN
• 1. Identity-Aware GNN
• ...and 2 more

TRANS (TRANS)

Count: 12 concepts (4.0%)

Concepts:

• 1. Graph Transformer
• 1. Graphormer
• 1. SAN (Spectral Attention Net)
• 1. GPS (General Powerful GNN)
• 1. Laplacian Positional Encoding
• 1. Random Walk Struct Encoding
• 1. Sign-Invariant Network
• 1. Basis-Invariant Network
• 1. Spatial Bias (Graph Attention)
• 1. Centrality Encoding
• 1. Edge Encoding (Transformer)
• 1. Relative Positional Encoding

SCALE (SCALE)

Count: 10 concepts (3.3%)

Concepts:

• 1. Mini-Batch Training (GNN)
• 1. Neighbor Sampling
• 1. GraphSAINT
• 1. Cluster-GCN
• 1. LADIES Sampler
• 1. Layer-Wise Sampling
• 1. Historical Embeddings (SIGN)
• 1. Graph Partitioning
• 1. Graph Sampling Strategy
• 1. SIGN Architecture

TOOLS (TOOLS)

Count: 9 concepts (3.0%)

Concepts:

• 1. PyTorch Geometric (PyG)
• 1. NetworkX
• 1. Open Graph Benchmark
• 1. SNAP
• 1. DeepSNAP
• 1. DGL
• 1. TUDataset
• 1. PyKEEN
• 1. RelBench (Benchmark)

HETERO (HETERO)

Count: 8 concepts (2.7%)

Concepts:

• 1. Heterogeneous GNN
• 1. R-GCN (Relational GCN)
• 1. Heterogeneous Graph Trans
• 1. HAN (Hetero Attention Net)
• 1. Meta-Path
• 1. Relation-Specific Weights
• 1. Basis Decomposition (R-GCN)
• 1. Type-Specific Projection

GEN (GEN)

Count: 7 concepts (2.3%)

Concepts:

• 1. Graph Generative Model
• 1. GraphRNN
• 1. GCPN
• 1. Molecule Generation
• 1. Variational Autoencoder (VGAE)
• 1. DiGress
• 1. Graph Generation Metrics

Recommendations

• ✅ Excellent balance: Categories are evenly distributed (spread: 12.3%)
• ✅ MISC category minimal: Good categorization specificity

Educational Use Recommendations

• Use taxonomy categories for color-coding in graph visualizations
• Design curriculum modules based on taxonomy groupings
• Create filtered views for focused learning paths
• Use categories for assessment organization
• Enable navigation by topic area in interactive tools

---

Report generated by learning-graph-reports/taxonomy_distribution.py