福建电脑2025，Vol.41Issue(9)：30-36,7.DOI:10.16707/j.cnki.fjpc.2025.09.006
中文文本情感分析系统研究与实现
Research and Implementation of a Chinese Text Sentiment Analysis System
刘文华 1邓友 1任保金2
作者信息
- 1. 重庆对外经贸学院重庆超大城市数字化治理学院 重庆 401520
- 2. 北京华晟经世信息技术有限公司 北京 101100
- 折叠
摘要
Abstract
To improve the accuracy and computational efficiency of sentiment analysis in Chinese texts,this study designed a multi method comparative analysis framework.The ChnSentiCorp Chinese sentiment dataset based on the Kaggle platform was used to vectorize text using TF-IDF feature engineering and BERT pre trained models,and the performance of Scikit learn,Spark MLlib,and BERT models was systematically compared.The experimental results show that the F1 score of the BERT model reaches 95%,but the computational cost is relatively high;SVM in Scikit learn has the best efficiency among traditional methods;Although Spark MLlib supports distributed training,it is susceptible to memory limitations in standalone mode.Verified the feasibility of Spark distributed computing in large-scale Chinese sentiment analysis.关键词
机器学习/深度学习/情感分析/中文文本/并行计算Key words
Machine Learning/Deep Learning/Sentiment Analysis/Chinese Text/Parallel Computing分类
天文与地球科学引用本文复制引用
刘文华,邓友,任保金..中文文本情感分析系统研究与实现[J].福建电脑,2025,41(9):30-36,7.
