| 曾寰,李金忠,付青.基于集合运算特征提取及Stacking策略的新闻多分类方法[J].井冈山大学自然版,2021,42(2):70-75 |
| 基于集合运算特征提取及Stacking策略的新闻多分类方法 |
| A MULTI-CLASSIFICATION METHOD OF NEWS CLASSIFICATION METHOD OF NEWS BASED ON SET OPERATION FEATURE EXTRACTION ON STACKING STRATEGY |
| 投稿时间：2020-12-07 修订日期：2020-12-28 |
| DOI：10.3969/j.issn.1674-8085.2021.02.012 |
| 中文关键词: 文本分类 新闻 集合运算 Stacking策略 |
| 英文关键词: text classification news set operation stacking strategy |
| 基金项目:国家自然科学基金项目（61762052）；江西省教育厅科技计划项目（GJJ180574）；江西省高校人文社科项目（JC19235） |
|
| 摘要点击次数: 1763 |
| 全文下载次数: 2703 |
| 中文摘要: |
| 文本分类是机器学习重要任务之一，如何对文本信息进行有效分类组织，对用户查找并获取有用信息具有重要作用。针对新闻文本分析，提出了一种基于集合运算特征提取及Stacking策略的新闻多分类方法，该方法基于集合运算的方法来提取文本特征，采用Stacking策略，使用SVM以及贝叶斯方法来对文本进行分类。与典型同类方法对比，在复旦大学文本分类数据集上的实验结果表明，该方法随着样本数增加，各分类指标逐渐升高并趋于稳定。 |
| 英文摘要: |
| Text classification is one of the important tasks in machine learning. How to classify and organize text information effectively plays an important role in user information retrieval. For news text analysis, a multi-classification method of news based on set operation feature extraction and stacking strategy was proposed. The method, firstly uses set operation to extract text features, then uses stacking strategy, SVM and Bayesian method to classify the text. Compared with the typical similar methods, the experimental results on text classification data set of Fudan University show that, with the increase of sample size, the classification indexes of this method gradually increase and tend to be stable. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
