Cinema Experiments 2015

Currently, two movie audience VOC studies have been completed: the first in 2013 measuring audiences across 108<br> sessions covering 16 movies and a second in 2015 across 85 sessions of 3 movies. These studies found that VOC<br> data from specific movies contained patterns that could be seen across multiple sessions of that movie.The first peak occurred when the main character’s dress caught fire, and the second occurred at the onset of the final battle scene. Furthermore, compared what was happening on screen with future VOC data and created backward prediction models that could determine what type of scene had recently occurred based on the audience’s subsequent VOC<br> data. Further studies tested whether VOC patterns were related to other elements, such as the age of the audience<br> and the movies’ age certification.目前已完成两项电影观众语音反馈（Voice of Customer, VOC）研究：第一项研究于2013年开展，覆盖108场观影活动，涉及16部电影；第二项研究于2015年开展，覆盖85场观影活动，涉及3部电影。两项研究均发现，特定影片的观众语音反馈数据中存在可在该影片多场观影活动中重复观测到的共性模式。该模式存在两个峰值：第一个峰值出现在主角衣物起火的场景，第二个峰值则出现在最终决战场景的开场时刻。此外，研究人员将影片画面内容与后续采集的观众语音反馈数据进行比对，并构建了反向预测模型：该模型可基于观众后续产生的语音反馈数据，反推当前刚刚发生的场景类型。后续研究进一步验证了观众语音反馈模式与其他因素的相关性，例如观众年龄以及电影的年龄分级认证。