1.TextCNN
1.1相關代碼下載
https://github.com/dennybritz/cnn-text-classification-tf
https://github.com/gaussic/text-classification-cnn-rnn
1.2 講解
https://blog.csdn.net/weixin_36604953/article/details/78324834
https://zhuanlan.zhihu.com/p/25928551
https://hunto.github.io/nlp/2018/03/29/TextCNN文本分類詳解.html
https://zhuanlan.zhihu.com/p/40276005
https://www.jianshu.com/p/f69e8a306862
https://blog.csdn.net/u010223750/article/details/51437854
1.3 https://github.com/gaussic/text-classification-cnn-rnn中相關函數說明
(1)learn.preprocessing.VocabularyProcessor
實現的功能就是,根據所有已分詞好的文本建立好一個詞典,然後找出每個詞在詞典中對應的索引,不足長度或者不存在的詞補0
https://blog.csdn.net/The_lastest/article/details/81771723
(2)kr.preprocessing.sequence.pad_sequences
將多個序列截斷或補齊爲相同長度。
https://keras.io/zh/preprocessing/sequence/
2.TextRNN
2.1TextRNN講解
https://zhuanlan.zhihu.com/p/28196873
https://zhuanlan.zhihu.com/p/44424550
2.2RNN講解
https://cloud.tencent.com/developer/article/1011162