We study the community detection problem by embedding the nodes of a graph into a n-dimensional space such that similar nodes remain close in their representations. There are many state-of-The-Art methods, like node2vec and DeepWalk to compute node embeddings with the use of second order random walks. These techniques borrow methods like the Skip-Gram model, used in the domain of Natural Language Processing (NLP) to compute word embeddings. This paper explores the idea of porting the GloVe (Global Vectors for Word Representation) model, a popular technique for word embeddings, to a new method called GloVeNoR, to compute node embeddings in a graph, and creating a corpus with the use of second order random walks. We evaluate the model's quality by comparing it against node2vec and DeepWalk on the problem of community detection on five different data sets. We observe that GloVeNoR discovers similar or better communities than the other existing models on all the datasets based on the modularity score.
- clustering,
- Community detection,
- global vectors,
- graphs,
- node representation,
- random walks,
- word embeddings
Available at: http://works.bepress.com/aikaterini-potika/47/