Sentiment Analysis Of Short Text Based On Cloud Environment

In recent years,with the rapid development of mobile Internet,social networks have been gradually integrated into people's daily life,with the result of unprecedentedly quick and convenient ways in which people publish,acquire and transmit information,thereafter,with the emergence of a large amount of subjective short text information.Such short text information mainly concentrates on reviews about hot events,and those of commodities,public opinions or personal views.Commonly,these short texts are sparsely populated,real-time,colloquial,and ambiguous.The sentiment analyses of the short texts,especially those from the Internet,are more and more concerned by the researchers.The sentiment analyses on the short texts in traditional algorithm fail to deal with the massive data in the Internet in real time.However,the emergence of large data parallelization technology makes it available to have a sentiment analysis on the short texts of massive data.In this thesis,based on the characteristics of short texts,firstly,the researchers adopt the SVM short text sentiment classification model in the stand-alone environment.In the experiment,six types of symbols have been used,namely,expression symbols,word clustering symbols,together with those of part-of-speech,n-gram,negation,and the sentiment dictionary.Secondly,with the applicance of model of the sentiment classification in short texts based on neural network,several neural networks have been adopted in the experiment,among which are convolutional neural networks,long and short memory networks convolutional memory networks.The experimental results show that the classification made by the two kinds of models is effective on short texts for sentiment analyses in the stand-alone environment.Finally,combined with cloud computing technology,with the algorithm of parallel improvement in the stand-alone environment,it is able to construct a modal of a sentiment analysis for the massive short texts with the foundation of cloud environment,which is able to deal well with the sentiment analysis on the massive short texts.With the assurance of classification accuracy,this model achieves a better implementation efficiency and scalability.