基于BiLSTM和CNN的序贯三支情感分类模型研究

赵梦宇, 孙京博, 魏遵天, 辛现伟, 宋继华

Research on sequential three⁃way sentiment classification model based on BiLSTM and CNN

Mengyu Zhao, Jingbo Sun, Zuntian Wei, Xianwei Xin, Jihua Song

表4 三个数据集在不同模型下分类F1对比

Table 4 Classification F1 scores of three datasets with different models

模型 Hotel Computer 中文微博情感 BiLCNN⁃S3WD 0.9145±0.0047 0.9234±0.0051 0.8213±0.0044 BiLSTM 0.9127±0.0045 0.9102±0.0085 0.8034±0.0100 BiLSTM+CNN(kernel_size=2) 0.9120±0.0033 0.9188±0.0030 0.7687±0.0208 BiLSTM+CNN(kernel_size=3) 0.9110±0.0020 0.9206±0.0054 0.8057±0.0118 BiLSTM+CNN(kernel_size=4) 0.9135±0.0045 0.9184±0.0093 0.8010±0.0153 BiLSTM+CNN(kernel_size=5) 0.9125±0.0040 0.9182±0.0057 0.8153±0.0105 DLSTWSC 0.9100±0.0000 0.9278±0.0000 0.7882±0.0229 CFRT 0.8100±0.0028 0.8254±0.0029 0.6737±0.0020