摘要
数据融合是运用系统诸多方面数据,采集目标对象或者环境中信息进行数据综合处理,从而获取数据中隐藏有效信息。在云计算环境下,针对分散文本数据提出基于阈值优化聚类的神经网络数据集成融合改进方法。运用k-近邻距离对分散文本数据对象实行顺序排列,采用分位数求解阈值并区别不同文本数据密度各个子序列,寻找与子序列相对应的局部阈值,同时对数据子序列进行聚类处理,利用数据结构储存每个簇。采用自组织映射对聚类文本数据进行离散化处理,根据文本数据实际要求,选择揭示离散数据属性的粗糙集衡量指标,通过粗糙集数据分析获知分散文本数据决策属性,将数据输入神经网络中根据判别规则进行文本数据融合。实验证明,运用文中方法可有效抽取各分散文本数据中存在的有价值数据信息。
Data fusion is the use of many aspects of the system data,collecting target objects or information in the environment to process data comprehensively,so as to get effective information hidden in the data.In the cloud computing environment,an improved method for data integration and integration of neural networks based on threshold optimization clustering is proposed for decentralized text data.The use of k-in order to implement the nearest neighbor distance dispersion text data,using quantile for threshold and different density of each sub sequence of text data,find the local threshold corresponding to the sequence,and the clustering of data processing sub sequence,using the data structure stored in each cluster.The self-organizing map is used to discretize the clustering of text data,text data according to the actual requirements,selection of discrete data attribute of rough set of indicators,through the analysis of the data obtained scattered text data attribute of rough set,the data input of neural network according to the judgement rules for text data fusion.The experiment proves that the use of the method can effectively extract the valuable data information in the scattered text data.
作者
娄道国
李若斌
刘冰
张冬冬
Lou Daoguo;Li RuoBin;Liu Bing;Zhang Dongdong(Dalian University of Technology,Dalian 116000,China;State Grid Dalian Electric Power Supply Company,Dalian 116000,China)
出处
《科技通报》
2019年第2期85-88,93,共5页
Bulletin of Science and Technology
关键词
云计算
文本数据
集成聚类
数据融合
cloud computing
text data
integrated clustering
data fusion
作者简介
娄道国(1980-),男,辽宁庄河人,高级工程师,研究方向:信息技术、大连理工大学、计算机科学与技术专业。
