基于局部敏感哈希算法的语音文档主题分类  

Speech document theme Panels based on local sensitive Hash algorithm

作　　者：何学文[1] 张磊[1] 

机构地区：[1]哈尔滨工程大学信号与信息通信工程学院,哈尔滨150001

出　　处：《大众科技》2012年第3期16-17,共2页Popular Science & Technology

摘　　要：对识别后的语音文档进行了向量空间模型的建立,针对得到的高维稀疏矩阵提出了基于局部敏感哈希的语音文档分类算法,算法能够直接在高维稀疏矩阵上进行分类,无需降维。此外,在构建局部敏感哈希函数的时候结合了稳定分布。实验证明,局部敏感哈希算法能够对语音文档进行合理有效的分类,同时获得了较小的时间复杂度。The vector space model was established after speech document was identified,according to be geted high dimension sparse matrix,sorting algorithm of local sensitive hash speech document was proposed,the algorithm can be directly classified on the high dimension sparse matrix and no dimension reduction.In addition,the paper combined with stable distribution when the local sensitive hash function was constructed.Experiments show that local sensitive hash algorithm can reasonably classified speech document,and while receive a small time complexity.

关 键 词：语音文档分类 局部敏感哈希 稳定分布 向量空间模型 

分 类 号：TN912.3[电子电信—通信与信息系统]