2026-10-14T11:02:50Z https://www.scidb.cn/oai

10.57760/sciencedb.j00104.00157 2026-09-10T10:52:02Z

2026-09-10 CMU-MOSI doi:10.57760/sciencedb.j00104.00157 en Originated from the YouTube platform, it contains a total of 93 video clips focusing on movie reviews, where all speeches and comments are presented in English by 89 different speakers. These videos are divided into 2199 clips, each of which carries a specific sentiment polarity and is manually labeled with sentiment labels ranging from -3 to +3, with a total of 7 levels. There are 7 levels of emotion labels from -3 to +3, with -3 indicating strong negative emotion, 0 indicating a neutral non-emotional state, and +3 being strong positive emotion. From -3 to 0, the degree of negative emotion gradually decreases; from 0 to 3, the degree of positive emotion gradually increases. This dataset provides rich and accurate data support for sentiment analysis research. CMU-MOSI; Multimodal Sentiment Analysis; YouTube Movie Reviews Yonghong Xie PUBLIC https://creativecommons.org/publicdomain/zero/1.0/ dataset Science Data Bank