A Product Concept Generation and Evaluation Method Based on Knowledge

Conceptual design is an important stage in product design, and function design is the basis of conceptual design. But the product that rely on the designer's experience always exists preferences and defects because of the limitations of designer's knowledge. To solve this problem, this paper proposed a product concept generation and evaluation method based on knowledge. Firstly, we adopted syntactic dependency parsing (SDP) to extract the main semantics structures from the existing product in the market. Secondly, word embedding was adopted for the vectorization of the main function structure, then a product base and a function base was constructed. Normally, every kind of product has its basic types of functions, so we used K-means to classify the functions. We then randomly choose a function from each cluster to combine a new product. As for the evaluation of each product, we calculated quantitatively the novelty, diversity and feasibility of solutions by semantic similarity. According to the comparison of the solutions and the existed products in novelty, diversity and feasibility. We can deduce that our methodology is feasible.