摘要
由于卷积神经网络在训练前就需要确定分类个数,所以用一个卷积神经网络模型能为文章添加的标签总个数是确定的不能改变的,而使用多个卷积神经网络模型组合可以避免这类问题。在实现多标签中首先使用词向量库将文字转化为词向量,然后依次用多个卷积神经网络模型对文章进行提取特征并分类,其中通过对预测出的类别概率分析来添加相应的标签,可以代替人工操作节省撰写文章人的时间。
Since the convolutional neural network needs to determine the number of classifications before training,the total number of labels that can be added to the article by a convolutional neural network model is determined and cannot be changed,and multiple convolutional neural network models are used.Combinations can avoid such problems.In the implementation of multi-labels,the word vector library is first used to convert the words into word vectors,and then the articles are extracted and classified by using multiple convolutional neural network models in turn,and corresponding labels are added by predicting the category probability analysis.Instead of manual operations,this can save time writing articles.
出处
《工业控制计算机》
2019年第12期55-56,共2页
Industrial Control Computer
关键词
卷积神经网络
文本分类
多标签
博客
convolutional neural network
text classification
multi-label
blog
