一种基于联合深度神经网络的食品安全信息情感分类模型.docx

一种基于联合深度神经网络的食品安全 信息情感分类模型 刘金硕张智 武汉大学计算机学院 摘要： 针对因屮文食品安全文本特征表达困难，而造成语义信息缺失进而导致分类器准 确率低下的问题，提出一种基于深度神经网络的跨文本粒度情感分类模型。以食 品安全新闻报道为目标语料，采用无监督的浅层神经网络初始化文本的词语级词 向量。引入递归神经网络，将预训练好的词向量作为下层递归神经网络 (Recursive Neural Network)的输入层，计算得到具备词语间语义关联性的句子 特征向量及句子级的情感倾向输出，同时动态反馈调节词向量特征，使其更加接 近食品安全特定领域内真实的语义表达。然后，将递归神经网络输岀的句子向量 以时序逻辑作为上层循环神经网络(Recurrent Neural Network)的输入，进一步 捕获句子结构的上下文语义关联信息，实现篇章级的情感倾向性分析任务。实验 结果表明，联合深度模型在食品安全新闻报道的情感分类任务中具有良好的效果, 其分类准确率和F1值分别达到了 86. 7%和85. 9%,较基于词袋思想的SVM模型有 显著的提升。 关键词： 联合神经网络模型;多粒度文本特征;词向量;食品安全;情感倾向性分析; 作者简介：刘金硕(1973-),女，博士，副教授,主要研究方向为模式识别、自然语 言处理、数字图像处理;E-mail: liujinshuo@whu. cdu. cn; 作者简介：张智(1990-),男，硕士，主要研究方向为自然语言处 理,E-mail:zhangzhi@whu. edu. cn(通信作者)。 收稿日期：2015-11-26 基金：国家自然科学基金资助 Sentiment Analysis on Food Safety News Using Joint Deep Neural Net work Model LIU Jin-shuo ZHANG Zhi Computer School, Wuhan University; Abstract： Facing the difficulties in feature expression of Chinese food safety text information, and the loss of semantic information with low classification accuracy, a sentimental text classification model based on joint deep neural network was presented. The proposed model utilizes the corpora of food safcty document captured from the internet，and word vector from word embedding method as the input for the neural network to get the pre-trained word vector. The pretrained word vector is further trained dynamically to get the word features and the sentimentai classification of the sentence result, which better express the phrase-level sentimental relations for each sent once and the real semantic meaning in the food safety domain. Then the word feature of the sentence is inputted to the recurrent neural network(RNN)to catch the semantic information of the sentence structure further,realizing the sentimental classification of the text.The experime nts show that our joint deep neural n etwork model achieves bet ter results in sentiment analysis on food safety information, compared with the bag-of-words based SVM model. The class