Large-scale network mining and analysis is key to revealing the underlying dynamics of networks. Lately, there has been a fast-growing interest in learning random walk-based low-dimensional continuous representations of networks. While these methods perform well, they can only operate on static networks. In this paper, we propose a random-walk based method for learning representations of evolving networks. The key idea of our approach is to maintain a set of random walks that are consistently valid with respect to the updated network topology. This way we are able to continuously learn a new mapping function from the new network to the existing low-dimension network representation. A thorough experimental evaluation is performed that demonstrates that our method is both accurate and fast, for a varying range of conditions.
展开▼
