基于文本聚类的档案数据全自动分类方法研究

Research on automatic classification of archival data based on Text Clustering

大数据时代下,档案文本数据规模海量递增,增加了聚类分析的难度,如何准确、高效的实现档案数据全自动分类。而针对以往PSI算法在高纬度、稀疏特征中易于陷入局部最优,本文引入GWO算法,通过对种群精英个体的克隆和变异,及基于DE算法的个体更新算法创新,来规避早期收敛和局部最优的问题,而后,引入FastText有监督快速分类方法,基于softmax分层技术完成海量档案数据的全自动化分类,以提升档案分类的可靠性和效率。

In the era of big data,the scale of file text data increases massively,which increases the difficulty of clustering analysis.How to realize the automatic classification of archival data accurately and efficiently.In order to avoid the problems of early convergence and local optimization,the GWO algorithm is introduced to avoid the problems of early convergence and local optimization by cloning and mutation of elite individuals in the population,and the innovation of individual update algorithm based on DE algorithm.Then,fasttext supervised fast classification method is introduced,In order to improve the reliability and efficiency of file classification,the full automatic classification of massive archive data is completed based on softmax layered technology.