Open Access

表3

本文方法与传统方法在Agnews和SST1上的性能比较

模型种类 模型 Agnews SST1
改进前F1 改进后F1 改进前F1 改进后F1
RNN-based bi-GRU 85.82±0.23 89.66±0.13 51.96±0.01 52.35±0.15
RNN-Capsule 84.54±0.24 88.96±0.52 51.31±0.06 51.62±0.36
bi-LSTM 85.54±0.18 89.76±0.15 47.64±0.26 51.01±0.50
CNN-based CNN 83.83±0.18 86.68±0.07 51.21±0.08 51.55±0.27
Text-CNN 82.69±0.12 87.24±0.32 50.01±0.49 50.25±0.97
Attention-based HAN 84.19±0.02 88.35±0.12 51.71±0.21 51.85±0.97
Transformer 84.26±0.31 89.25±0.02 50.31±0.11 51.25±0.12

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.