Patent data-driven product functional innovation based on bipartite network

Traditional market research relies on experts’ personal experience or small-scale survey data, which makes it difficult to provide effective and reasonable decision-making support for product design. To improve product design efficiency at the early design stage, a patent data-driven product functional innovation (PFI) method is proposed based on a bipartite network. The function knowledge mined from patent data is applied to construct the corresponding bipartite network, which is used to calculate the weights of patent nodes and function nodes. Subsequently, a random walk algorithm (RWA) is introduced to calculate the convergence probability between function nodes and is combined with a restart algorithm (RA) to assign nodes’ weight. Ultimately, six PFI strategies based on corresponding weight combination methods are proposed to reconfigure existing product functions. A case study on the multifunctional shower head shows that this method can effectively obtain functional optimisation schemes and provide new avenues for product development research.