基于Keras框架的文本情感分析

Text sentiment analysis based on Keras framework

随着社交媒体日新月异地发展,我们实际上生活在一个巨大的社交媒体网络中,每一个用户或者说每一个社交账号都是这个庞大网络中的一个节点。每个节点都与多个节点连接,信息流在由这些节点构成的网络上进行流动。这些节点之间有的相互交换信息,有的作为中间传输节点进行信息的转发。作为网络上进行传输的信息也是各式各样,如图片数据,视频数据以及文本数据。其中,文章数据作为最根本的信息流动在网络上。这些文本数据背后隐藏的用户的情感或者观点进行分析,将能够为社交媒体研发者提供强有力的反馈信息,进而完善社交平台的开发以产生更有价值的利润。文章基于深度学习框架Keras实现对淘宝上的用户评价信息的情感分析。主要进行了两个模型的实验:Keras-bert模型和RNN(LSTM)网络,实现对评论数据的情感分析。

With the rapid development of social media,we actually live in a huge social media network,and every user or every social account is a node in this huge network.Each node is connected to multiple nodes,and the information flow flows on a network composed of these nodes.Some of these nodes exchange information with each other,and some act as intermediate transmission nodes to forward information.As the information transmitted on the network is also various,such as picture data,video data and text data.Among them,text data flows on the network as the most fundamental information.Then analyzing the emotions or opinions of users hidden behind these text data will be able to provide strong feedback for social media developers,and then improve the development of social platforms to generate more valuable profits.This article is based on the deep learning framework Keras to achieve sentiment analysis of user evaluation information on Taobao.Two experiments were mainly performed:Keras-bert model and RNN(LSTM) network to realize sentiment analysis of review data.