【摘要】 为了获得利益,在线评论当中有很多伪造的评论,酒店领域的评论也不例外。由于酒店领域的评论只有少部分的标注数据,这给深度学习技术的应用带来了困难。提出一种融合酒店消费领域的专业知识和基于文本卷积神经网络的方法。该方法对数据进行增强,用无监督学习方法获得aspect情感信息,利用卷积神经网络识别虚假评论。实验结果表明,该方法的识别效果比传统方法有显著的提升。更多还原

【Abstract】 Too many forged comments online has result in an issue that customer aren’t able to distinguish it. The comments on hotel field have only a small part of dimension data, which makes it difficult to apply the deep learning technology. This paper proposes a method that combines professional knowledge in the hotel field and a text-based convolutional neural network. It enhanced the data, then used unsupervised learning method to obtain aspect emotional information, and used convolution neural network to identify false comments. The experimental results show that the recognition effects of this method have a significant improvement compared with the traditional method.更多还原