Abstract: The variable precision Bayesian rough set(VPBRS)approach is a flexible method for Kansei knowledge acquisition to accommodate the individual differences within a user group. In order to handle the possible combinatorial explosion at the stage of Kansei rule extraction， an improved algorithm based on sequential covering strategy is proposed. Basically， the approximation regions of Kansei decision classes are taken as the input， and the selection of conjunctive items with maximum covering ability is taken as the greedy search strategy for rule specialization. On this basis， the approximation region is covered step by step through iterative learning， and the decision rule set is extracted. A basic example and a design example of toaster appearance are conducted， whose results show that the improved VPBRS approach is effective.

Key words: Kansei engineering, knowledge acquisition, decision rule, Bayesian rough sets, sequential covering strategy