Construction Of ConvSLSTM Model For Learners’ English Vocabulary Knowledge Level In Adaptive Learning System

Based on a variety of deep learning technologies,this study attempts to integrate CV machine vision,NLP natural language processing and SNN spiking neural network technology,apply these technologies to English vocabulary adaptive learning system,and construct learners’ knowledge level model（ConvSLSTM learners’ knowledge level model）.The model is an integration and improvement of ConvSLSTM model and Spiking LSTM model.It performs well in performance indexes such as time-consuming,accuracy and convergence speed.The article consists of six chapters.Chapter 1 Introduction.Firstly,it briefly introduces the research background,including the era background and research motivation.The value of adaptive learning is discussed in the context of the times.The research motivation includes the need to make use of the advantages of spiking neural network compared with traditional deep learning,the necessity of the application of deep knowledge tracking technology,and the research results that the author’s research group has achieved in the early stage.Then it states the theoretical and practical significance,research objectives,research methods and research contents.Chapter 2 Research review.It mainly combs the achievements and shortcomings of the application of in-depth knowledge tracking technology in adaptive learning system;This chapter combs the achievements of the construction of learners’ knowledge level model based on deep learning technology（including Text Cnn,RNN,LSTM,On LSTM,Text Rcnn,Easy Rcn,DPRcnn,Conv GRU,etc.）and the existing defects,and discusses the possible methods to improve the defects.Chapter 3 is the construction of ConvSLSTM learner knowledge level model.It mainly states the description,establishment and solution process of ConvSLSTM learner knowledge level model.Firstly,we briefly introduce and design ConvSLSTM learner knowledge level model,which is an important basis of Conv LSTM.Then,the modeling process of ConvSLSTM learner knowledge level model is described in detail,including three stages: problem description,establishment and solution.The problem description introduces the basic unit of Spiking LSTM and the mathematical formula of the basic unit,establishes the design covering data coding format,basic neuron layer,network structure and forward propagation,and finally writes Python implementation code based on Pytorch framework to solve it.Chapter 4 is the test of ConvSLSTM learners’ knowledge level model.Through a series of comparative experiments between models,compare the index training time-consuming,average accuracy,average accuracy per unit time,accuracy change,loss loss change and so on,so as to test the advantages of ConvSLSTM learner knowledge level model.Chapter 5 is the application of ConvSLSTM learner knowledge level model.The ConvSLSTM learner knowledge level model is applied to English vocabulary learning practice to verify the effectiveness of the model in real education scenes.Chapter 6 is the research conclusion and the prospect of follow-up research.This paper summarizes the results of this study,and dissects the defects and possible solutions of ConvSLSTM learner knowledge level model,as well as the possibility of further research.