Research On Text Analysis Method For The Social Evaluation Of Personal Training Quality In Higher Education

With the vigorous development of education and the constant emphasis on the quality of personnel training in China,the era of big data in "data driven schools,analysis changed education" has arrived.Following the rapid development of the mobile Internet,a large number of evaluation data about the personnel training quality in higher education are generated in the network.These data are of great value to the study of how to improve the quality of personnel training.Therefore,it is urgent to use text analysis technology to effectively capture and reasonably analyze these data,so as to realize the automatic collection and analysis of target data.So as to provide decision-making support for relevant departments in personnel training.To this end,this thesis proposes to construct a text analysis method for the social evaluation of personnel training quality in higher education,the main research contents are as follows:(1)According to the different structure of Internet social media websites,develop a multi-source data collection scheme,design web crawlers to crawl target data.On this foundation,further construct the index system of personnel training quality in higher education and build standard data sets.(2)Sentiment analysis is conducted on the data sets of social evaluation of personnel training quality in higher education.Considering the network environment of the data source and the domain of the data,social network dictionary and the domain dictionary are added to the basic emotion dictionary.So we formulate a calculation method of text emotion intensity based on improved emotion dictionary.(3)Text classification is carried out on the data sets of social evaluation of personnel training quality in higher education.Considering the semantic features and context relevance of text data,the Recurrent Neural Network is used to solve context semantic relevance problems and Attention Model is used to assign higher weights to domain keywords.Thus,We combine the recurrent structure and attention model in order to build a text classification model based on bidirectional recurrent attention neural network.In summary,firstly,this thesis designs and completes the collection and processing of multi-source data for target data.Secondly,we propose the calculation method of text emotion intensity based on improved emotion dictionary and the text classification model based on bidirectional recurrent attention neural network to analyze target data.Finally,the prototype system is designed to show the analysis results.