| 引用本文: | 程圣军,黄庆成,刘家锋,唐降龙.一种改进的ML-kNN多标记文档分类方法[J].哈尔滨工业大学学报,2013,45(11):45.DOI:10.11918/j.issn.0367-6234.2013.11.008 |
| CHENG Shengjun,HUANG Qingcheng,LIU Jiafeng,TANG Xianglong
.An improved ML-kNN approach for multi-label text categorization[J].Journal of Harbin Institute of Technology,2013,45(11):45.DOI:10.11918/j.issn.0367-6234.2013.11.008 |
|
| 摘要: |
| 针对应用传统k近邻算法进行多标记文档分类时忽略了标记之间相关性的问题,提出了一种改进的ML-kNN多标记文档分类方法．针对文本特征的特点,采用一种基于KL散度的距离尺度来更好地描述文档相似度．根据近邻样本所属类别的统计信息,通过一种模糊最大化后验概率法则来推理未标记文档的标记集合．与ML-kNN不同的是,该方法可以有效地利用标记相关性来提升分类性能．在3个标准数据集上,5个多标记学习常用评测指标下的实验结果表明:所提方法在多标记文档分类问题上要明显优于ML-kNN、Rank-SVM和BoosTexter等主流多标记学习算法． |
| 关键词: 文档分类 多标记学习 标记相关性 k近邻 KL散度 |
| DOI：10.11918/j.issn.0367-6234.2013.11.008 |
| 分类号: |
| 基金项目:国家自然科学基金资助项目(7,8); 黑龙江省自然科学基金资助项目(F201021). |
|
| An improved ML-kNN approach for multi-label text categorization |
|
CHENG Shengjun, HUANG Qingcheng, LIU Jiafeng, TANG Xianglong
|
|
(School of Computer Science and Technology, Harbin Institute of Technology, 150001 Harbin, China)
|
| Abstract: |
| Conventional kNN algorithms ignore label correlations when being applied to multi-label text categorization. To cover this shortage, an improved Multi-label kNN approach for text categorization is proposed. A specific distance metric based on KL divergence is derived to measure the similarity between individual documents. Based on statistical information gained from the label sets of neighboring documents, a fuzzy maximum a posteriori principle is utilized to conjecture the label sets of the unlabeled documents. Different from ML-kNN, the proposed approach can exploit label correlations to improve classification performance effectively. Experiments on three benchmark datasets using 5 popular multi-label evaluation metrics suggest that the proposed approach achieves superior performance to some well-established multi-label learning algorithms, such as ML-kNN、Rank-SVM and BoosTexter. |
| Key words: text categorization multi-label learning label correlations k nearest neighbor KL divergence
|
