Research On The Analysis Model Of Classroom Teaching Behavior Based On LSTM

Scientific and effective teaching behaviors in classroom teaching have an important impact on the quality of classroom teaching.Using certain techniques and methods to observe and analyze classroom teaching behaviors helps teachers comprehensively and accurately understand classroom dynamics,reflect on teaching behaviors,and improve teachers' professional ability and classroom teaching quality.The traditional Flanders interaction analysis method observes and quantifies the language interaction behavior in classroom teaching,and uses the interaction analysis matrix to analyze classroom teaching behavior,although it can simplify the complexity of the classroom teaching and research process,there are problems such as incomplete classification of behavior categories and low efficiency of manual analysis.Therefore,optimizing and improving traditional classroom teaching behavior analysis methods and exploring classroom teaching behavior analysis methods in the new era are of great significance to overcome the limitations of traditional classroom teaching behavior analysis methods and analyze classroom teaching behaviors in depth.Based on the traditional Flanders interaction analysis method and the Long Short Memory Network(LSTM)in deep learning technology,this paper proposes a new classroom teaching behavior analysis method.This method first optimizes the Flanders coding system in the coding stage to build a classroom teaching behavior dataset.Then in the analysis phase,based on the LSTM time series classification model,an LSTM-based classroom teaching behavior analysis model is constructed to classify the classroom teaching behavior time series,and the classroom teaching is analyzed and evaluated based on the classification results.Finally,through comparative experiments,the feasibility and effectiveness of the method are verified.The specific research contents are as follows:(1)Based on the Flanders interactive analysis method,this article first revises and improves the classroom teaching behavior classification and coding standard system based on modern information technology links,and refines the classroom teaching behavior categories to better reflect the integrity of the classroom teaching process,richness and diversity.Then develop classroom teaching evaluation indicators,design and develop classroom teaching behavior coding and labeling software.Using a combination of manual observation and software assistance,the classroom teaching behaviors are coded and labeled at a certain sampling interval to build a classroom teaching behavior dataset and lay a data foundation for the subsequent experiments.(2)According to the characteristics of classroom teaching behavior coding data and the research characteristics of classroom teaching behavior analysis methods,this paper builds on the LSTM time series classification model based on the LSTM time series classification model.Then the model is trained based on the classroom teaching behavior dataset,and the classification accuracy is used as the evaluation index of the model.Use the trained model to classify the classroom teaching behavior coding data without classifying classroom teaching evaluation index categories,and complete the analysis and evaluation of classroom teaching behavior according to the specific content corresponding to the class label in the classification result.(3)In order to verify the feasibility and effectiveness of the classroom teaching behavior analysis model based on LSTM,this article uses Flanders interaction analysis method and LSTM-based classroom teaching behavior analysis method to analyze and evaluate the same classroom teaching video.Then,the analysis results of the two analysis methods are compared from four aspects:analysis dimension,analysis time,analysis accuracy,and analysis scale.Based on the analysis results,the feasibility and effectiveness of the LSTM-based classroom teaching behavior analysis method are verified.The experimental results show that the classroom teaching behavior analysis method based on LSTM can not only realize the basic analysis function of Flanders interactive analysis method,but also further improve the analysis function and expand the analysis dimension.At the same time,the involvement of coding software and deep learning technology greatly shortens the coding time and analysis time,improves the analysis efficiency and accuracy,and makes the analysis results more helpful for teachers to reflect on teaching behaviors,thereby improving teaching quality.