Knowledge graph with machine learning for product design

Abstract

Knowledge graph is a particular form of graph that represents knowledge through entities and relations. Machine learning, particularly deep learning, can be adopted to construct, interpret, and enrich knowledge graph towards unknown entities and relations. As a knowledge-intensive endeavour, product design can greatly benefit from knowledge graph with machine learning. A structured framework is proposed to develop design-specific knowledge graph, based on which, deep learning is leveraged to learn graph embeddings, make predictions, and support reasoning. The framework effectiveness is validated through a quantitative experiment, where knowledge graph is used to make design-related predictions about smart products in home environment.