微信息大数据粗糙集的近似约简

为了应对微信息舆情数据的格式复杂、价值稀疏和收集困难等大数据处理技术难题,基于隐含语义分析和粗糙集近似约简理论,设计微信息的数据区间值集和近似匹配分类算法.在不影响数据主要关联关系的原则下,提炼核心属性、消减次要属性,实现一种微信息异常主题倾向的发现方法.结果表明,该近似约简算法能在完成微信息兴趣倾向主题分类的前提下,将数据集属性大幅度缩减,提高微信息的信息挖掘效率,为微信息大数据舆情处理工作提供了新的思路和案例.

Approximate reduction of micro message big data rough set

In order to deal with such technological problems in big data processing as complex format, sparse value and difficult collection of micro message public opinion data, based on the latent semantic analysis （LSA） and rough set approximate reduction theory, the data interval value set and approximate matching classification algorithm of micro message were designed. Under the principle of not affecting the main association relationship of data, the core attributes were extracted, the secondary attributes were reduced, and a method of discovering the micro message abnormal theme tendency was realized. The results show that under the premise of completing the classification of micro message interest tendency themes, the proposed approximate reduction algorithm can greatly reduce the data set properties, improve the information mining efficiency of micro message, and provide a new thought and case for the processing work of public opinion of micro message big data.