关键词：
Product innovative design
Functional basis
Machine learning
Patent evaluation
TRIZ
摘要：
The reuse of design knowledge within a specific domain is currently a major research focus in product design. However, the importance of support from cross-domain knowledge is steadily on the increase for product innovative design, which includes multi-domain knowledge recombination, transfer and transformation. As the biggest knowledge provider, patent information plays an irreplaceable role inspiring designers in the conceptual design stage. Nevertheless, it is of a great importance and also an issue to figure out how to efficiently retrieve cross-domain patents. In this paper, we propose a function-based patent knowledge retrieval tool for conceptual design of innovative products. In order to establish the underlying local database, function information, technical terms and International Patent Classification (IPC) information are respectively extracted from patents crawled on Websites to represent function, technology and domain properties, thereby generating a patent knowledge space. Particularly, by using a semi-supervised learning algorithm, function information is automatically classified and labeled by functional basis, into which the design problem is abstracted and interpreted forming a design problem space. By functional basis mapping from the design problem space to the patent knowledge space, we retrieve the required cross-domain patents, which are then clustered and evaluated in accordance with the extracted technology and domain properties. The selected patents are thus purposefully recommended to trigger designers' creativity. Finally, a case study illustrates the conceptual design process based on the proposed retrieval tool. In addition to that, we conducted an experiment to verify our tool. It demonstrates that our proposed tool can assist designers to generate more ideas and the novelty of ideas is higher. (C) 2019 Elsevier B.V. All rights reserved.
欢迎来到三峡大学图书馆！
