MindSpore【课程作业经验】基于TextCNN文本情感分类

#基于TextCNN文本情感分类

在本次实验中我们用mindspore实现TextCNN的针对aclImdb情感分类任务。

数据加载

在这里基于TextCNN的方法我们需要指定文本句子长度，并且对句子进行处理（留下字母和空格其余符号删去）标定标签：pos：1，neg：0

maxlen =20
sentences =[]
labels=[]
posdirname = "aclImdb\\train\\pos\\"
negdirname = "aclImdb\\train\\neg\\"
file_num =10000
for txtfile in os.listdir(posdirname)[:file_num]:
    newline=""
    with open(posdirname+txtfile,encoding="utf-8") as txt:
        line = txt.read()
        s = ''.join(ch for ch in line if (ch.isalnum()|ch.isspace()))
        sentences.append(s[:maxlen])
        labels.append(1)
        
for txtfile in os.listdir(negdirname)[:file_num]:
    newline=""
    with open(negdirname+txtfile,encoding="utf-8") as txt:
        line = txt.read()
        s = ''.j