图书情报领域多源数据特征级融合方法研究综述

A Review of the Multi-source Data Feature-level Fusion Method in the Field of Library and Information Science

[目的/意义]多源数据特征级融合是将多源或多维数据中描述研究对象的多层次特征进行融合的方法手段,用于揭示数据内在的关联性与互补性,从而提供更全面、更精确的分析视角。梳理相关研究对于丰富科研人员研究手段、促进情报研究的智能化具有重要意义。[方法/过程]在阐述特征级融合研究的重要性和必要性的基础上,梳理图书情报领域特征级融合方法近10年的研究成果。[结果/结论]目前图书情报领域所运用的多源数据特征级融合方法可归纳为基于线性加权的融合方法、基于耦合的融合方法以及基于神经网络模型的融合方法三类。同时总结各种特征级融合方法的适应场景以及图书情报领域在使用这些方法中所暴露的问题,并结合学术环境展望未来的发展路径。

[Purpose/Significance]Multi-source data feature-level fusion is a method devised for the extraction and integration of multi-level features from diverse and multidimensional data sources,which is used to reveal the intrinsic correlations and complementarities of the data.This provides a more comprehensive and precise analytical perspective.The review of relevant research is of great significance to enrich research methods and promote the intelligent process of information science research.[Method/Process]On the basis of elaborating the importance and necessity of feature-level fusion research,this paper compiled the research results of feature-level fusion methods in the field of library and information science in the past 10 years.[Result/Conclusion]At present,the multisource data feature-level fusion methods used in the field of library and information science can be summarized into three categories:linear weighted fusion methods,feature-coupled fusion methods and neural network model based fusion methods.It demonstrates the applicable scenarios of various feature-level fusion methods and analyzes the problems exposed in the use of these methods in the field of library and information science,and prospects its future development in the academic environment.