Abstract
This paper discusses the critical issue in product design which is related to the semantics corpora of current successful products. In order to establish a reliable model based on the semantic merits of products, and to provide an extensible learning database as a reference for new-product development, we propose an approach to construct semantic space based on PLSA (Probabilistic Latent Semantics Analysis). We devise the whole process and then use chair as the examined product in our case, construct the semantic space of its form from scratch. Afterwards, we discuss several scenarios that the established semantic space could be applied to in the activity of product development.