基于集成算法的在线课程评论情感识别与主题挖掘研究

On Online CourseReview Emotion Recognition and Topic Mining Based on Integrated Algorithm

本文构建了集成算法对在线课程评论进行多元分类情感识别与主题挖掘的模型,通过Adaboost多个弱分类器之间的相互加权组合成强分类器对课程评论进行三分类情感识别,提取到不同特征样本下的在线课程评论情感的分类结果,并运用LDA主题模型挖掘评论的隐藏主题,最后搭建语义网络,帮助学习者从整体把握课程的优劣属性及关注主题。以MOOC平台的10583条评论为对象,进行情感识别模型构建,并与机器学习的单独分类模型进行实验对比。实验结果发现,该模型能够有效地识别评论主体的情感,准确率优于单独的分类器,准确值高达88.12%,并能较好地抽取评论关注主题及课程属性,为学习者在选择课程时提供策略支持,帮助学习者做出正确决策,这说明集成学习算法在帮助学习者挑选课程做决策的性能上适应度较高。

The study builds a multi-classified emotion recognition and theme mining model of Adaboost integrated algorithm for online course reviews,and uses the mutual weighted combination of multiple weak classifiers into a strong classifier to carry out three-classified emotion recognition of course reviews,and extracts them from samples of different characteristics.The online course comments on the classification results of emotions,and uses the LDA theme model to explore the hidden themes of comments,and finally builds a semantic network to help learners grasp the advantages and disadvantages of the course and pay attention to the theme as a whole.The study takes 10583 comments on the MOOC platform as the object,builds an emotion recognition model and compares it with the separate classification model of machine learning.The experimental results found that this model can effectively identify the emotions of comment subjects,with better accuracy than a separate classifier,with an accuracy value of 88.12%,and can better extract class comments to focus on topics and course attributes to help learners provide strategic support when choosing courses to make correct decisions,which shows that integrated learning The algorithm is highly adaptable in the performance of helping learners choose courses and make decisions.