欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对中文问题分类方法中特征向量维数过高导致处理速度过慢的问题,提出一种基于局部鉴别索引和支持向量聚类的中文问题分类方法。首先利用局部鉴别索引算法对原始高维问句数据集进行降维,将其映射到一个低维空间中,然后通过支持向量聚类算法对问句进行分类。在哈工大社会计算与信息检索研究中心的中文问题集上进行实验,实验结果证明了该方法的有效性,大类准确率87.6%,小类准确率72.5%,取得了较好的效果。
英文摘要：
Aiming at the problem of jogging speed resulted from too high dimensions of the eigenvector in Chinese question classification,we put forward a Chinese question classification method which is based on locality discriminating indexing（LDI）and support vector clustering（SVC）. First,the LDI algorithm is used to reduce the dimensions of the original high dimensional question dataset and the question dataset is then mapped onto a low dimensional space,subsequently the questions are classified by SVC algorithm. The experiment has been made on the Chinese question set of the research centre of social computing and information retrieval at HIT. Experimental results prove the effectiveness of the method,the accuracy of the coarse classes and the fine classes achieve 87. 6% and 72. 5% respectively. The experiment achieves pretty good results.
同期刊论文项目
同项目期刊论文
期刊信息
位置：