Driving risk evaluation is a critical task for traffic safety and vehicle insurance. Unfortunately, many existing methods disregard the unknown and uncertain nature of driving behavior in real-world scenarios. In this paper, we formula this task as a multi-criteria group decision-making problem and propose a dual perspective information volume method for evaluating the driving risk score based on linguistic Z-number. Our proposed method accurately captures the uncertainties and reliability of driving behavior by utilizing two fuzzy numbers A and B. The main contributions of our paper are as follows: firstly, we introduce the α-truncation set and the information volume-based utility function and credibility probability model, which form the basis of the vehicle evaluation method. Secondly, we consider cross-information to model the divergence of information between dual perspectives, based on different experts and features. We calculate the credibility value for each perspective using the Z+-based information volume. Thirdly, we present the dual perspective calculation method used to assess driving risk. This method assigns risk scores to individual vehicles based on a utility function. Finally, we conduct a case study using the historical trajectories of vehicles obtained from an insurance company. The results are compared against the damages coefficient. Furthermore, we perform dependability and parameter analysis to demonstrate the effectiveness and superiority of our approach. Our approach provides valuable insights and suggestions for the insurance industry and traffic safety departments in relevant fields.