基于Attention-CLSTM模型的商品评论分类

Classification of Commodity Reviews Based on Attention-CLSTM Model

文本分类是自然语言处理中的一项重要基础任务,指对文本集按照一定的分类体系或标准进行自动分类标记。目前网络文化监督力度不够、不当言论不受限制,导致垃圾评论影响用户体验。因此提出一种基于注意力机制的CLSTM混合神经网络模型,该模型可以快速有效地区分正常评论与垃圾评论。将传统机器学习SVM模型和深度学习LSTM模型进行对比实验,结果发现,混合模型可在时间复杂度上选择最短时间,同时引入相当少的噪声,最大化地提取上下文信息,大幅提高评论短文本分类效率。对比单模型分类结果,基于注意力机制的CLSTM混合神经网络模型在准确率和召回率上均有提高。

Text classification is an important basic task in natural language processing,which automatically classifies text sets according to certain classification systems or standards.Problems prevail because of current lack of supervision of the network culture and restriction of improper comments,which leads to the problem of spam comments affecting the user experience.In this paper,a CLSTM hybrid neural network model based on attention mechanism is used to classify commodity reviews,which can quickly and effectively distinguish normal comments from spam comments.At the same time,compared with the traditional machine learning SVM model and the deep learning LSTM model,it is found that the hybrid model can select the shortest time in time complexity,introduce relatively little noise,and maximize the extraction of context information in the comments.The effect of short text classification has been significantly improved.Compared with the single model classification results,the CLSTM used in this paper has improved both accuracy and recall rate.