A bottleneck in high-throughput nanomaterials discovery is the pace at which new materials can be structurally characterized. Although current machine learning (ML) methods show promise for the automated processing of electron diffraction patterns (DPs), they fail in high-throughput experiments where DPs are collected from crystals with random orientations. Inspired by the human reasoning process from multiple perspectives (i.e., DPs of a crystal taken from different zone axes), we modeled automated crystal system identification from DPs with arbitrary zone axes as a sequential decision-making process by multiview opinion fusion. The location and intensity information of the Bragg disks in DP images are extracted and used as inputs for ML. We designed a custom CNN to capture the physical parameters embedded with the DPs by working off the adjacency relations between Bragg disks. The CNN was trained in the context of evidential deep learning, which provides opinions (classification probabilities) and quantifies uncertainties. The decision is made through the fusing of the opinions from multiple zone axes and is automated through the guidance of the quantified uncertainty. Our framework achieves high testing accuracy (0.94) and is shown to be robust under noisy scenarios involving Bragg disk perturbation, vacancy and redundancy. Finally, we have demonstrated the framework on real experimental data with high accuracy. In the context of experimental materials discovery, where data can now be generated faster than it can be analyzed, this work sets the stage for integrating ML frameworks into high-throughput experimental workflows, not only delivering the much-needed processing power, but also providing a basis for autonomous decision-making in the materials discovery pipeline.