Theory and Applications of Graph Neural Networks in Knowledge Acquisition

Abstract

Knowledge acquisition is a process in which an artificial intelligence system acquires relevant knowledge from datasets. Knowledge acquisition extracts and forms knowledge from data of different structures. One type of data used to acquire knowledge is structured data, such as physics models, chemical structures, social network information, and traffic network information. Many structured data can be represented as graph structures with dependencies and attributes. In recent years, deep learning methods are applied to knowledge acquisition tasks. As a deep connection model, graph neural network (GNN) is suitable for processing data represented as graph structure since it can use nodes and edges to realize information interaction and obtain graph dependencies. In this review, we discuss GNN model and its variants. Experiments show that the variants of GNNs outperform traditional methods in many knowledge acquisition tasks.