Situational Awareness In Sentiment Analysis

With the rapid development of the Internet,artificial intelligence has gain widespread concern.Its further study breakthrough into a new stage of development.Under the background,as one closely related discipline sentiment analysis's relevant research work have also been expanded.The core of the article is:?analysis existing text sentiment analysis method,compare the effect of a variety of emotional classification trained by traditional machine learning models;? introduce ensemble learning methods,use random forest as meta learning method train base classifiers which trained through different feature sets,propose a new model"MFMB-ME,Multi-Features Multi-Base-Classifiers Meta Ensemble Learning Sentiment Analysis Model".Though the experiments concluded that:by using a different set of features and different base classifiers,the ensemble model can obtain significant promotion;?through a combination of text sentiment analysis and situational awareness,put forward the text sentiment analysis situational awareness model:"SA-SA,Sentiment Analysis Based On Situational Awareness Model".The paper introduces the development of artificial intelligence and the significance and value of text sentiment analysis,analysis the current development situation and problems.Introduce the text sentiment analysis's relevant popular technology,including "emotion dictionary based method" and "machine learning method".Introduce situational awareness methodology.Secondly,compare the difference of the result of text sentiment classification by traditional machine learning,a single feature set,multi-feature sets of meta-learning multiple classifiers ensemble learning.Experiments of traditional machine learning use decision trees,support vector machines,logistic regression and other methods,also compare results of classification performance by different traditions classification machine learning methods;use random forests as ensemble learning method train classifier based on a single feature set and analysis the classification performance;multi-features-classifiers meta ensemble learning method use the different combination of different text feature set(including word,stem,part of speech,grammar,ngram etc.)and different base classifier(logistic regression,language models,etc.)train classifier by random forest as meta-learning method the integrated,analysis the classification performance by different combination strategies.Finally draw conclusions to the results of the experiment and research,this paper proposes the engineering reality emotion classification model:MFMB-ME,the model's feature is able to consider a variety of natural language processing feature set obtain high classification accuracy via an ensemble approach.Its classify result is better than the traditional model and the single feature set ensemble model.