基于“问题——方法”知识元挖掘的学科知识流动研究

Research on Subject Knowledge Flow Based on“Problem-Method”Knowledge Units Mining

[目的/意义]研究问题和研究方法构成一个学科的主要研究特征,知识元是知识内容的最小单元,通过对学术文献中表达研究问题的知识元和表达研究方法的知识元进行抽取和关系识别,有助于刻画学科的细粒度知识流动特征,同时对科研创新起到促进作用。[方法/过程]通过学术文献中问题知识元和方法知识元的自动化抽取,综合采用深度学习算法、词向量和词共现结合方法、层次聚类算法、空间距离计算方法等多种手段识别知识元间的三元组关系,从而构建学科内部细粒度知识流动网络,在此基础上对知识流动特征进行总结和阐述。[结果/结论]在CNKI中选取图书情报领域的中文核心期刊论文数据,根据本文研究模型,从论文摘要中识别问题知识元和方法知识元及知识元间的关系,构建图书情报领域的“问题—方法”知识流动网络,基于此展开实证分析,验证了模型的可行性和有效性。

[Purpose/Significance]Research problems and research methods constitute the main characteristics of a subject.Knowledge unit is the smallest elements of knowledge content.This paper mines the knowledge units that express research problems and research methods in academic literature.It is helpful to describe the fine-grained knowledge flow characteristics of the subject and promote scientific innovation at the same time.[Method/Process]Through the automatic extraction of problem knowledge units and method knowledge units from academic literature,it identifies the triplet relationship among knowledge units by a comprehensive combination of deep learning algorithm,word vector and word co-occurrence combination,hierarchical clustering algorithm,spatial distance calculation and other methods,so as to build the fine-grained knowledge flow network within the subject and summarize its characteristics.[Result/Conclusion]From the data of Chinese core journals in library and information in CNKI,it identifies the problem knowledge units,method knowledge units and their relationship from the abstract of the paper,and constructs the"problem-method"knowledge flow network.By this empirical analysis,it verifies the feasibility and effectiveness of the model.