基于CNN-BiGRU的商品短文本分类算法研究

Short Text Classification Algorithm Based on CNN-BiGRU

研究商品短文本分类,有利于推动商品流通和实现个性化商品信息推送。为克服独立卷积神经网络(CNN)仅提取文本局部信息的不足,提出CNN和双向门控循环单元(BiGRU)相结合的文本分类模型。该模型不仅提取了文本局部特征,还有效利用了上下文信息,最后采用A-Softmax函数优化分类效果。实验结果表明,基于CNN-BiGRU模型在商品短文本分类上的各个评价指标均优于其他对比模型。

The research on the classification of short text of goods is conducive to promoting the circulation of goods and realizing the push of personalized commodity information.In order to overcome the shortage of independent CNN that only extracts local information of text,this paper proposes a text classification model that combines CNN and BiGRU.This model not only extracts the local features of the text,but also effectively uses the context information.Finally,A-Softmax function is used to optimize the classification effect.The experimental results show that each evaluation index based on CNN BiGRU model in short text classification is superior to other comparison models.