Sentiment Analysis Of Micro-blog Text Based On New Word Discovery

With the rapid development of Internet technology and mobile applications,the amount of network information is increasing day by day.As an open Internet social platform,micro-blog is increasingly favored by huge number of users and play a powerful role in public opinion expression and transmission,because of its characteristics of convenience,user anonymity and information disclosure.Nowadays,micro-blog has become an important channel for people to share new information and express their personal feelings.There are a huge amount of short text messages shared in the micro-blog platform every day,and anyone who published a micro-blog message is likely to spread rapidly throughout the network.A large number of subjective text data on micro-blog is growing with exponential speed,in order to dig the real emotional preferences of people accurately,the sentiment analysis in micro-blog short text has become a hot research area in the field of Natural Language Processing.With the appearance of all kinds of network popular new-words at the same time,to find the real attitude of public from the massive short text data has important research value: on the one hand,it can help enterprises find real product reputation to make some improvements,on the other hand,it can also help the government or relevant departments to take some actions towards hot events.At present,the development of Chinese emotion classification task is limited in the field of Natural Language Processing,and many reserchers are committed to studying the model or algorithm to improve the accuracy of sentiment classification.According to the existing research,this paper propose a method of new word discovery based on statistics and a model of sentiment classification method based on rule set,which can adapt to the sentiment oriented classification task of micro-blog short text.The experimental results show that the method we proposed can improve the performance of classifier effectively.First of all,this paper summarized a large number of literatures in related fields,and analyzed the methods of text emotion recognition and classification.And compared the sentiment classification methods based on machine learning and rule set,and concluded the characteristics of micro-blog text.Secondly,combined with the characteristics of Chinese micro-blog,this paper proposed the new word discovery method based on statistics and the sentiment classification model based on rule set.One is to build a text sentiment dictionary based on the existing research to adapt micro-blog text,which is play an importantrole in the classification of emotional words in the Internet;another is to design the rule set of text sentiment in word level and sentence level,in order to improve the accuracy of emotional value calculation.Finally,this paper uses the large-scale micro-blog text data set crawled on Sina micro-blog platform to accomplish the new word discovery experiment and emotion classification experiment,and compares the performance of our method and the traditional method in the emotional tendency classification task by experiment standards.Experiments show that the method proposed in this paper can improve the accuracy of Chinese micro-blog sentiment classification effectively.