Modelling the Dynamics of Customer Requirements Considering Their Lability and Sensitivity in Product Development.

Enterprises are challenged with fulfilling continuously changing customer requirements (CRs) that are influenced by many uncertain factors. Existing research only modelled the dynamics of CRs (DoCRs) by focusing on the potential resultant effects of CR changes, while lacking insightful exploration of the change characteristics of CRs themselves. To fill this gap, an online review-based DoCRs modelling method is proposed considering the lability and sensitivity of CRs by analyzing the changes in customers' perceptions of product features over time. First, CRs in online reviews are extracted through Latent Dirichlet Allocation. Second, Bidirectional Long Short-Term Memory is applied to measure customer sentiments in different periods towards specific product features. The probabilistic linguistic term sets are employed to characterize customer sentiment distributions which are then incorporated into the modelling of DoCRs. Third, a novel CRs classification cube is established by integrating the lability, sensitivity, and importance of CRs from an evolution-adapted perspective. Finally, the DoCRs of a smartphone is investigated as a case study to demonstrate and effectiveness of the proposed methodology. Several managerial implications for the selection and adjustment of product development strategy are provided.