A Study of Knowledge Discovery and Pattern Recognition Based on Large-Scale Sentiment Data in Online Education for College Students

With the development of internet, online education platforms have become main means acquiring knowledge for people. To improve efficiency user analysis, this paper proposes a framework sentiment recognition based on course review information platforms. Firstly, according to characteristics data platforms, authors use crawlers and open-source word separation tools complete collection collation; secondly, TextCNN (text convolutional neural networks) BILSTM (bi-long-short-term-memory) combine build feature layer fusion text classification model. The results show that average precision model built is 95.1% three classifications. Finally, test reviews new courses conducted. rate all types exceeds 90%. Therefore, proposed provides ideas analysis.