Latent semantic indexing for capitalizing experience in inventive design

Abstract

The growing complexity of the design activity in an innovation and sustainable context requires experience reuse as a means to limit unsustainable investments. It is a crucial task for both academic and industrial communities to find ways to efficiently capture and reuse past experience. Case-based reasoning (CBR) is a research paradigm that stores experience as a knowledge unit to solve a new problem from the previous design experience. A well-established method for inventive design is IDM (the Inventive Design Methodology). Its most widely used tool to solve a problem is the “Contradiction Matrix” associated with forty inventive principles. The correct use of these tools needs the mapping from freely expressed text (Specific Parameters or SPs) into a well-established set of Generic Engineering Parameters (or GEPs). This mapping requires expertise and may, if inappropriately used, lead to weak results. This paper introduces the Latent Semantic Indexing (LSI) algorithm to discover the implied semantic relations between SPs and GEPs coming from past experience. A semantic space based on the LSI results is built for guiding retrieval in case-based reasoning.