Fusion of Text and Graph Information for Machine Learning Problems on Graphs

Nowadays, an increased attention is drawn towards Network Representation Learning, which is a technique that maps nodes of a network into vectors of a low dimensional space, preserving their similarity. The vectors can later be used for different downstream tasks such as Node Classification, Link Prediction and Graph Visualization. Naturally, some networks have text information associated with them. For instance, in a citation network each node is a scientific paper associated with its abstract or title, in a social network all users might be viewed as nodes of a network and posts of each user as textual attributes. This work studies how combination of existing methods can increase accuracy on the downstream tasks and proposes some modifications to popular architectures to be able to better capture textual information. Finally, new dataset is mined, which can be used for evaluation of the Link Prediction task.