Skip to content

References: Machine Learning and Graph ML

  1. Graph Neural Network - Wikipedia - Overview of neural network architectures that operate on graph-structured data including GCN, GraphSAGE, and GAT. Explains message-passing frameworks for learning node and graph representations from organizational data.

  2. Node Embedding - Wikipedia - Techniques for learning low-dimensional vector representations of graph nodes (Node2Vec, DeepWalk, LINE) that preserve network structure. Used for downstream tasks like employee classification and link prediction.

  3. Link Prediction - Wikipedia - Methods for predicting missing or future connections in a network based on existing structure. Applicable to predicting future collaborations, mentoring matches, and organizational relationship formation.

  4. Network Science - Albert-László Barabási - Cambridge University Press (2016) - Comprehensive textbook on network science covering scale-free networks, preferential attachment, network dynamics, and spreading phenomena. Provides theoretical grounding for understanding organizational graph patterns.

  5. Graph Representation Learning - William L. Hamilton - Morgan & Claypool Publishers (2020) - Focused treatment of graph ML methods including node embeddings, graph neural networks, and knowledge graph completion. Directly applicable to building ML models on organizational graph data.

  6. Supervised Learning - Wikipedia - Overview of classification and regression methods trained on labeled data. Applicable to predicting employee outcomes (flight risk, performance, promotion) from graph features.

  7. Unsupervised Learning - Wikipedia - Clustering, dimensionality reduction, and anomaly detection methods that discover structure without labels. Applicable to finding natural employee groupings and detecting unusual network patterns.

  8. Bias in Machine Learning - Wikipedia - Examines sources of bias in ML systems including training data bias, measurement bias, and feedback loops. Critical for ensuring organizational analytics models do not perpetuate discrimination.

  9. PyTorch Geometric - PyG Team - Documentation for the graph neural network library built on PyTorch, providing implementations of GCN, GAT, GraphSAGE, and other architectures used in graph ML research and applications.

  10. Feature Engineering - Wikipedia - Process of creating informative input variables from raw data for ML models. In organizational analytics, graph-derived features (centrality scores, community membership, path lengths) serve as powerful predictors.