Research On Analysis And Application Of Collaborative Learning Discourse Based On Text Mining

In recent years,collaborative learning has been widely used in education and teaching to encourage students to create shared knowledge.Interactive dialogue is the basis of collaborative learning.It can be said that the process of collaborative learning is the process of discourse.The discourse text generated in the process of collaborative learning carries the main content of collaborative learning.Analyzing the discourse text can clarify the rule and essence of collaborative learning,and it is the key to mastering the changes in the process of collaborative learning.However,based on the literature review of relevant studies,it is found that the current collaborative learning discourse analysis research has not yet formed a sound system,and there are still the status quo of imperfect discourse analysis theory,lack of discourse analysis framework and lack of automated analysis tools.In this context,the research explores discourse analysis of collaborative learning,taking“framework design-model construction-application analysis” as the main line of research,tries to combine text mining to analyze collaborative learning discourses,and build a multidimensional discourse analysis framework to explore how to use text mining technology to automatically analyze collaborative learning discourses and measure the effect of collaborative learning.Discourse analysis is implemented from the theoretical exploration level to the practical application level,to improve the efficiency of collaborative learning process analysis,provide process evaluation basis for collaborative learning,and promote the innovation of collaborative learning practical application.In the framework design period,the collaborative learning group is taken as the research unit,starting from the organizational structure of the collaborative learning discourse text,the vocabulary and sentences are clearly regarded as the analysis level of the collaborative learning discourse,and the analysis elements of the two levels are determined.The vocabulary level includes word frequency statistics,correlation calculation,topic extraction,and vertical depth distribution,which are realized through Wordcloud,Bert,LDA,Matplotlib,and other technologies.The sentence level includes the total number of sentences,average sentence length,and discourse text classification,which are realized through Python statistics and machine learning.This stage mainly defines the corresponding relationship between the analysis elements and the collaborative learning effect,providing a clear objective basis for the evaluation of collaborative learning effect.In the model construction period,machine learning algorithm is used to classify the discourse text.The modeling steps include determining the classification criteria,text collection and annotation,text pre-processing,text vectorization,model training and selection.Firstly,the specific classification criteria for the three analytical dimensions of cognitive interaction,emotional interaction,and social interaction were determined;After text collection,annotation and preprocessing,vectorization of discourse text is realized by Word2Vec;Finally,model training was conducted using Decision Tree,Random Forest,Naive Bayes,and Support Vector Machine algorithms.The classification model with the best performance in text classification was selected based on evaluation criteria,including cognitive interactive discourse text classification model based on Random Forest,emotional interactive discourse text classification model based on SVM,and social interactive discourse text classification model based on Random Forest.The three classification models lay the technical foundation for automatic classification of discourse texts.In the application analysis period,the discourse analysis framework proposed in this study is applied to the actual collaborative learning session process,and based on the discourse analysis data obtained,the changes and differences of collaborative learning effects within and between groups are compared,which shows that the collaborative learning discourse analysis framework based on text mining has certain value and practicality in the actual collaborative learning discourse analysis application,which can provide an analytical basis for measuring the effect of collaborative learning.Then,based on the results of data analysis,the discourse process feedback and evaluation enlightenment of collaborative learning is summarized.Finally,based on this framework,the functional design and architecture design of collaborative learning intelligent discourse analysis tool are proposed,aiming to provide objective and powerful support for subsequent collaborative learning discourse analysis.To sum up,analysis and application research of collaborative learning discourse based on text mining has designed an automated multi-dimensional discourse analysis framework,which connects the relationship between text mining data and collaborative learning effects.It is hoped that the framework will provide a new way of thinking and analysis for subsequent collaborative learning discourse analysis,provide framework support for the development of collaborative learning intelligent discourse analysis tools,and enrich the path of intelligent analysis of collaborative learning discourses.