基于B-DMM的蒙医药文本数据聚类算法研究  

Research on Mongolian Medicine Text Data Clustering Algorithm Based on B-DMM

作　　者：包启明 裴志利[1] 姜明洋[1] 熊露 BAO Qi-ming;PEI Zhi-li;JIANG Ming-yang;XIONG Lu(College of Computer Science and Technology,Inner Mongolia Minzu University,Tongliao 028043,China;College of Mathematics and Physics,Inner Mongolia Minzu University,Tongliao 028043,China)

机构地区：[1]内蒙古民族大学计算机科学与技术学院,内蒙古通辽028043 [2]内蒙古民族大学数理学院,内蒙古通辽028043

出　　处：《内蒙古民族大学学报（自然科学版）》2023年第2期124-128,共5页Journal of Inner Mongolia Minzu University：Natural Sciences

基　　金：国家自然科学基金项目(62162049)。

摘　　要：为能够更好地了解蒙医药领域研究的热点和前沿,在GPU-DMM主题模型聚类算法的基础上构建一种基于背景主题的B-DMM聚类模型来获取蒙医药文本当中的隐含主题,给未来蒙医药相关研究提供参考。B-DMM模型利用外部知识以及优化生成过程来克服蒙医药短文本数据中存在的单词稀疏性问题。实验中采用主题语义一致性作为模型的评价标准,将结果与LDA、DMM、GPU-DMM、BTM等模型进行对比分析。实验结果表明,B-DMM模型在蒙医药热点主题挖掘方面的表现均优于基线算法,可以为相关研究者提供更有价值的信息。In order to better understand the hot spots and frontiers of research in the field of Mongolian medicine,a B-DMM clustering model based on background topics was constructed on the basis of GPU-DMM theme model clustering algorithm to obtain the implied themes in the text of Mongolian medicine,to provide reference for the future related research of Mongolian medicine.The B-DMM model used external knowledge and optimized the generation process to overcome the word sparsity problem in the text data.Subject semantic consistency was used as the evaluation criterion of the model,and the results were compared with LDA,DMM,GPU-DMM,BTM and other models.The experimental results show that the B-DMM model performs better than the baseline algorithm in mining hot topics of Mongolian medicine,which can provide more valuable information for relevant researchers.

关 键 词：B-DMM模型 GPU-DMM模型 蒙医药文本 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]