Design Concept Generation from Patent Information based on Novelty Potential and Distributional Word Representation Model

Abstract

It is essential to explore many new design concepts in the upstream phase of design to create innovative products that have potential competitive advantages in the market. This paper proposes a method to support the generation of new design concepts from intellectual property information. It adopts the theory of novelty potential that the more distant the two abstract concepts are, the more novel design concept the combination of them would generate. The distributed word representation model of word2vec measures the concept distant. This research extracts the abstract concepts from sentences of patent abstracts. The novelty potential matrix and the void matrix are used to choose an abstract concept combination that has a large novelty potential and no existing patent. As a case study, this research uses patent information of sports business with blockchain and generates concepts of new sports business models by the proposed method. The results verify its effectiveness.