The Platform Design And Implementation Of Text Sentiment Classification Based On Semi-supervised

The Internet has entered an era of data explosion,all kinds of people on the Internet expression the opinions of wide variety of events,items and people freely.The majority of these massive text data contain the expression of sentiment.The sentiment polarity classification of the text can provide decision support for network security,event prediction,shopping guide,public opinion analysis and so on.Who has the data in 21th century,who has the dominant power.Therefore,we designed a chapter-level text sentiment polarity classification platform that is mainly for the distinction and classification of commendatory and derogatory of a text,the main contents are as follows.(1)Method of text sentiment classification based on semi supervised learningThe accuracy of traditional text sentiment classification depends on the size of the training data and the quality of the annotation.It consumes a lot of manpower and time.Semi supervised machine learning method can be used as an effective means for automatic or semi-automatic expansion of training corpus.This paper introduce two kind of semi-supervised learning methods.The method includes self-training and active learning.SVM and ME to classify the unlabeled data.In this paper,we use the stepwise optimization classification model.Finally,a single model and integrated model are classified by using the optimized classification model.(2)Text sentiment classification and data analysis.For Chinese text sentiment classification,the Chinese data is chaotic,In particular,the user comment text,each person has a different expression.The platform has realized the function of data preprocessing.Including to stop word,text word segmentation,data of quantify.These functions are well prepared for the text sentiment polarity classification.This platform can be used as a text sentiment polarity classification system.The platform includes supervised text sentiment classification,semi supervised text sentiment polarity classification.The subsystem can also be used as an auxiliary data annotation system.(3)Realization of text sentiment classification platform This paper built a Java-based text sentiment classification polarity platform based on java.The platform can handle the data published by users in various websites.In order to accelerate the running speed of the platform.The parallel computation of Java is introduced in the platform.Users can choose different functions according to different needs.Each module can be used alone and Each module can be re-develop.