The paper presents a semi-supervised learning framework based on graph embeddings, which jointly predicts class labels and neighborhood context in a graph. The authors develop both transductive and inductive variants of the method. In the transductive variant, class labels are determined by learned embeddings and input feature vectors, while in the inductive variant, embeddings are defined as a parametric function of feature vectors, enabling predictions on unseen instances. The framework, named Planetoid, is evaluated on various benchmark tasks, including text classification, distantly supervised entity extraction, and entity classification, showing improved performance over existing models. The inductive method outperforms the second best inductive method by up to 18.7% in accuracy, and the best inductive and transductive methods outperform all other compared methods by up to 8.5% and 4.1% respectively. The paper also discusses related work in semi-supervised learning, graph-based semi-supervised learning, and embedding learning, and provides a detailed experimental setup and results.The paper presents a semi-supervised learning framework based on graph embeddings, which jointly predicts class labels and neighborhood context in a graph. The authors develop both transductive and inductive variants of the method. In the transductive variant, class labels are determined by learned embeddings and input feature vectors, while in the inductive variant, embeddings are defined as a parametric function of feature vectors, enabling predictions on unseen instances. The framework, named Planetoid, is evaluated on various benchmark tasks, including text classification, distantly supervised entity extraction, and entity classification, showing improved performance over existing models. The inductive method outperforms the second best inductive method by up to 18.7% in accuracy, and the best inductive and transductive methods outperform all other compared methods by up to 8.5% and 4.1% respectively. The paper also discusses related work in semi-supervised learning, graph-based semi-supervised learning, and embedding learning, and provides a detailed experimental setup and results.