Network Embedding Based on DepDist Contraction

Networks provide a very understandable representation of the data in which we get information about the relationships between pairs of nodes. For this representation, we can use one of the powerful analytical tools, 2D visualization. In visualization, networks have an alternative vector representation to which a wide range of machine learning methods can be applied. More generally, networks can be transformed to a low-dimensional space by network embedding methods. In this paper, we present a new embedding method that uses a non-symmetric dependency to find the distance between nodes and applies an iterative procedure to find a satisfactory distribution of nodes in space. Using experiments with small networks and dimension 2, we show the effectiveness of this method and discuss its properties.