基于PSO-LSTM的中文微博情感分类研究

Research on Sentiment Classification of Chinese Microblog Based on PSO-LSTM

微博短文本蕴含着较为丰富的情感信息,基于微博数据的情感分析已成为网络舆情监测的重要任务。为提高中文微博情感分类效果,提出一种基于粒子群优化(PSO)的长短期记忆(LSTM)模型(PSO-LSTM),该模型在LSTM模型的基础上进行了参数优化,能够更有效获取微博信息。实验以新冠肺炎疫情期间的微博数据集构建PSO-LSTM模型,与其它模型进行了比对实验。实验结果表明,PSO-LSTM模型能够有效提升中文微博情感分类的性能。

Microblog short texts contain rich emotional information.Sentiment analysis based on microblog data has become an important task for network public opinion monitoring.In order to improve the effect of sentiment classification on Chinese microblog,a long and short-term memory model(LSTM)was proposed,which was based on Particle Swarm Optimized(PSO).The parameters of this model were optimized on the basis of LSTM to obtain microblog information more effectively.The PSO-LSTM model was constructed from the micro-blog data set during the Covid 19 outbreak and compared with other models.The experimental results showed that the PSO-LSTM model can effectively improve the performance of Chinese microblog sentiment classification.