Text Orientation Analysis System Based On Neural Network In The Research And Implementation

With the Information age coming, millions of Internet users gradually form the network community whose representatives are news comments, chat rooms, blogs, wiki and micro blog. So a new phenomenon which is called public opinion can’t be ignored now.The main points of innovation of the essay are that it proposes the principal of training Orientation words, proposes Text Orientation Analysis Algorithm based on Neural Network, including the algorism and its derivative:Automatic Acquisition Algorithm of Orientation Words based on Neural Network. And it can help to improve Orientation Analysis both on words and articles.Researchers usually divide the comments into two kinds:positive or negative. There are two ways to do this:one is to use emotional knowledge and the other is based on classification. The former kind gets the orientation dictionary by knowledge database, and then analyzes the comments by calculating the weighting sums of the comments. Our algorithm is one of these algorithms. It updates the orientation dictionary by training it with Neural Network, thus can increases the accuracy of prediction. We also perform an experiment to prove that the system can achieve a better accuracy.The update of orientation dictionary is an important research project in Orientation Analysis. However the algorisms till now will somehow involve some manual work. To improve this we proposes Automatic Acquisition Algorithm of Orientation Words based on Neural Network, which trains candidate orientation words via Neural Network Training method and filter some words to get updated orientation dictionary. The paper describes the algorithm in detail. We perform an experiment to prove that the algorithm can acquire orientation words automatically and can achieve a better accuracy in Text Orientation Analysis.Finally we summarize our work and we look ahead the future research work. I proposed some suggestion for the future work; hope these will be helpful for later researchers.