In this project we will focus on both algorithms that create graph embeddings, applications of the graph embedding procedures and their evaluation. A graph embedding can be described as a mapping between a graph and a low dimensional abstract space which approximately preserves graph distances in the low dimensional space. Possible uses of online graph embeddings are numerous -- exemplar applications include link prediction, semi-supervised node labeling, community detection, graph visualization and routing on the network. A graph embedding procedure is considered to be inductive if it is able to map previously unseen nodes to this latent space based on their connections to existing nodes in the graph. Simply, this means that offline methods are unable to generalize to nodes that were unseen during training time. On account of this developing online methods is extremely relevant and they would be considerably useful in real world applications.