WCSE

A Multi-granularity Knowledge Representation and Mining Method for Patent Texts

Abstract— To support patent knowledge reuse for product concept design, multi-granularity representation of patent knowledge should be taken into consideration. Traditionally, an ontology model called TCO (Techspecs Concept Ontology) is used for representing the hierarchical architecture of a product, which consists of components layer, function modules layer, and product layer. However, the interactions such as the context relevance among components are ignored in the TCO model. In this paper, a modified model called PSG-TCO is proposed, where the PSG (Patent Semantic Graph) part can be used to capture components’ interactions. Based on the PSG-TCO model, an automatic knowledge extraction method is proposed to construct PSG-TCO instances from a large number of patent texts. All of the PSG-TCO instances form a multi-granularity knowledge base, which can be used for engineering knowledge retrieval, design concept discovery, and providing potential innovation stimulus.

Index Terms— knowledge-based design, patent analysis, PSG-TCO model.

Lin Gong
School of Mechanical Engineering, Beijing Institute of Technology, Beijing, China; Yangtze Delta Region Academy of Beijing Institute of Technology, Jiaxing, China
Mingren Zhu
School of Mechanical Engineering, Beijing Institute of Technology, Beijing, China
Zhenchong Mo
School of Mechanical Engineering, Beijing Institute of Technology, Beijing, China
Ziyao Huang
School of Mechanical Engineering, Beijing Institute of Technology, Beijing, China

[Download]

Cite: Lin Gong, Mingren Zhu, Zhenchong Mo, Ziyao Huang, "A Multi-granularity Knowledge Representation and Mining Method for Patent Texts," WCSE 2022 Spring Event: 2022 9th International Conference on Industrial Engineering and Applications, pp. 239-246, Sanya, China, April 15-18, 2022.