特征工程研究领域发展趋势的可视化分析

Visualization analysis of development trend of the feature engineering research

为了解国内外关于特征工程的研究进展与态势、研究热点与现状,使用CiteSpace软件对中国知网CNKI和Web of Science核心合集数据库收录的研究主题为特征工程的173篇中文文献和555篇外文文献进行计量学分析与可视化处理。通过对年度发文量、研究国家、研究作者、文献关键词等进行分析,得出结论:自2015年以来本研究领域中外文发文数量持续增长。中国和美国的相关研究成果居多,约占文献总数量的64%。LeCun Y、Bengio Y、刘挺、林鸿飞是较有影响力的研究作者。国内的研究热点主要有协议识别、xgboost、深度学习。国外的研究热点主要有深度学习、迁移学习、实体识别。研究前沿是深度学习和因子分解机。

In order to understand the progress,trend,hotspot and current status of feature engineering research at home and abroad,the CiteSpace software is used to analyze and visualize 173 Chinese literatures and 555 foreign literatures of feature engineering collected in the Core Collection Database of CNKI(China National Knowledge Infrastructure)and Web of Science.Based on the analysis of the annual volume,research country,researcher,and keyword,it is concluded that the number of Chinese and foreign publications in this research field has continued to increase since 2015.China and the United States have the most relevant research papers,accounting for about 64%of the total literature.LeCun Y,Bengio Y,Liu Ting and Lin Hongfei are influential researchers in this research field.Domestic research focuses on protocol recognition,xgboost and deep learning.Foreign research mainly focuses on deep learning,transfer learning and entity recognition.The research frontiers are deep learning and factorization machine.