多标签网页的Gauss-PNN粗糙集排序推荐
李又玲;常致全;杨浩
【期刊名称】《计算机应用研究》 【年(卷),期】2017(034)002
【摘要】为解决传统算法在网页多标签推荐过程中存在的信息不确定和较多结构冗余问题,提出基于高斯PNN粗糙集期望的多标签网页推荐算法.首先,基于粗糙集方法对标准概率数据流神经网络进行改进,提高其数据不确定处理能力;其次,为解决标准概率数据流神经网络数据覆盖性较差,且网络结构具有较大冗余,导致其无法快速识别新增标签的问题,基于附加的高斯块及其新增、组合及移除功能,对概率神经网络进行改进,解决标准PNN模型无法准确表达新增类别数据的问题,并对标签进行排序,实现新增数据的实时性预测;最后,利用所提算法对Yahoo多标签推荐实例进行验证,结果显示所提方法的推荐精度及效率更高.%In order to solve the problem of uncertain information and redundant information in the process of multi label recommendation of Web pages,this paper proposed a multi label Web recommendation algorithm based on Gauss-PNN rough set theory.Firstly,it improved the standard probabilistic data stream neural network with the method of rough
set,which
could
improve
the
ability
foe
the
data
processing.Secondly,in order to solve the problem of poor neural network data covering,it improved the probabilistic neural network based on the addition Gauss function block and its add,remove and combination function,which solved the problem of the standard PNN
多标签网页的Gauss-PNN粗糙集排序推荐
