分类法映射研究综述

A Review of Classification Mapping

[目的/意义]了解分类法映射相关研究,总结映射方法及应用场景,为相关研究人员提供参考。[方法/过程]对分类法映射基本概念进行描述,调研国内外相关研究,并简要阐述映射过程及应用场景;整理归纳已有分类法映射方法,并分析其局限性。[结果/结论]目前分类法映射主要有人工映射和自动映射两种模式,基于同现的映射、基于社会网络分析法的映射、基于交叉检索的映射、结合语义与结构的映射和基于机器学习的映射这5种方法。相比较而言自动映射各方法准确率较低,基本保持在60%左右,仍需优化自动映射算法以提高映射的准确率和覆盖率。

[ Purpose/significance ] This paper aims to understand the relevant research of classification mapping, summarize the mapping methods and application scenarios, and provide references for relevant researchers. [ Method/process ~ On one hand, the paper describes the basic concepts of classification mapping, surveys domestic and foreign research, and briefly explains the mapping process and application scenarios. On the other hand, the paper summarizes the existing classification mapping methods, and analyzes the limitations. [ Results/conclusions] At present, there are mainly two classification mapping modes： manual map- ping and automatic mapping, and five methods： co-occurrence mapping, mapping based on social network analysis, cross-retrieval- based mapping, mapping combining semantic and structure, and machine-learning mapping. The accuracy of automatic mapping method is relatively low, which remains at about 60%, so the automatic mapping algorithm still needs to be optimized to improve the mapping accuracy and coverage.