The Application of Graph Embedding Based on Random Walk

Abstract

In the historical process of scientific development, computers have a lofty position, and in recent years, graph embedding algorithms and models are one of the most popular subjects. A large number of similar data structures are indistinguishable by humans, but graph embedding can quickly compare and analyze these data structures. Existing research on random walk-based graph embedding methods is very rich. In order to summarize and classify the status quo of the more mature classical models and compare and integrate them, many different classical models are discussed in this paper. Based on different models, the problems solved, algorithm ideas, strategies, advantages, and disadvantages of the models are discussed in detail, and the application performance of some models is evaluated. DeepWalk model, Node2Vec model, HARP model are three graph embedding models based on the classical random walk model. Calculations for different data can occur by generating different node sequences. The three most important models in attribute random walk models are TriDNR model, GraphRNA model and FEATHER model. The model that only targets the information data in the shallow network is no longer suitable for the rapidly developing network. Attribute random walk models can handle data in deeper networks. At the end of this paper, the full text is summarized and the future prospect of this field is made.