基于数据共享的大数据特征快速提取方法

Fast feature extraction method of big data based on data sharing

随着信息化水平的提升,数据具有体量增加、来源增多等特性,特征提取速度与精度受到了不利影响,为数据共享带来了极大的阻碍,故提出基于数据共享的大数据特征快速提取方法研究。由于环境、设备、人为等因素的影响,导致获取的大数据中存在多噪、缺失等问题,以数据库文本数据为例,对其进行消噪、补全等处理。以此为基础,应用遗传算法选择大数据特征,使用节点部署模型,在数据共享方法下快速提取大数据特征。以提取特征为依据,采用线性分类器对数据进行分类,根据数据集的关联性构建数据共享模型,实现大数据特征快速提取。实验数据显示:相较于对比方法,应用提出方法获得的大数据特征提取时间更短,特征提取平均误差更小。

With the improvement of information level,data has the characteristics of increasing volume and sources.The speed and accuracy of feature extraction have been adversely affected,which has brought great obstacles to data sharing.Therefore,a fast feature extraction method of big data based on data sharing is proposed.Due to the influence of environment,equipment,human and other factors,there are many noise and missing problems in the obtained big data.Taking the text data of database as an example,noise elimination and completion are carried out.On this basis,the genetic algorithm is applied to select the big data features,and the node deployment model is used to quickly extract the big data features under the data sharing method.Based on the extracted features,the linear classifier is used to classify the data,and the data sharing model is constructed according to the relevance of the data set to realize the rapid extraction of big data features.The experimental data show that compared with the comparison method,the feature extraction time of big data obtained by the proposed method is shorter and the average error of feature extraction is smaller.