一种融合LSTM预测网络的试题并行推荐算法

A Parallel Recommendation Algorithm for Test Based on LSTM Prediction Network

设计了一种融合LSTM预测网络,基于多决策树和认知诊断的试题并行推荐算法MDT&CD-LSTM,用于解决自适应教育中的学习资源推荐问题。该算法集成多决策树和认知诊断的推荐结果,并利用LSTM网络预测学生的知识状态,从而向学生推荐最合适的试题。实验结果表明,该自动推荐算法可以有效地提高试题推荐结果的准确性,比改进决策树模型精确度提升了21.67%,误差均值减少了26.52%。该算法能够满足学生的个性化学习需求,帮助学生更好地理解知识点,具有广阔的应用前景。

This paper designs a parallel recommendation algorithm(MDT&CD-LSTM)based on multiple decision trees and cognitive diagnosis,which integrates LSTM prediction network,to solve the problem of learning resource recommendation in adaptive education.The algorithm integrates recommendation results of multiple decision trees and cognitive diagnosis,and uses LSTM network to predict students'knowledge state,so as to recommend the most suitable test questions to students.The experimental results show that the automatic recommendation algorithm can effectively improve the accuracy of test recommendation results,which is 21.67%higher than the improved decision tree model,and the mean error is 26.52%lower.The algorithm can meet students'personalized learning needs,help students better understand knowledge points,and has broad application prospects.