New China Ceramic Knowledge Clustering Based on Title Similarity
刘陶 1李锦 1邱望仁1
扫码查看
点击上方二维码区域，可以放大扫码查看
作者信息
1. 景德镇陶瓷大学
折叠
摘要
20世纪50年代至70年代,一些陶瓷研究机构、大专院校和专业瓷厂相继在景德镇成立,使得567陶瓷(又称新中国陶瓷)迈向数千年来瓷业的又一高峰.新中国陶瓷文献蕴含着丰富的款识、装饰手法、工艺流程、烧炼技术等陶瓷知识,是许多新中国陶瓷研究者和爱好者提高辨识和欣赏能力的原始依据.但由于新中国陶瓷文献在网络中难寻足迹,造成利用数据挖掘技术很难为所有新中国陶瓷知识进行归类整理,因而笔者以手里整理的新中国陶瓷新闻数据为数据集,尝试利用TF-IDF(term frequency-inverse document frequency)和BERT(Bidirectional Encoder Representations from Transformers)两种方法结合新中国陶瓷专业术语实现了基于标题相似度聚类新中国陶瓷知识的工作,为后期全面聚类新中国陶瓷知识提供参考,实验结果发现TF-IDF方法倾向以标题关键词聚类,而BERT方法则倾向以标题语义为聚类依据.
Abstract
From the 1950s to the 1970s,some ceramic research institutions,colleges,and professional porcelain factories were successively established in Jingdezhen,which made 567 Ceramics(also known as New China ceramics)reach another peak in the ceramic industry for thousands of years.The New China ceramic literature contains rich ceramic knowledge such as inscription,decoration techniques,process flow,and firing technology,which is the original basis for many New China ceramic researchers and enthusiasts to improve their identification and appreciation abilities.However,due to the difficulty in finding the footprints of New China ceramic literature on the Internet,it is difficult to classify and organize all New China ceramic knowledge using data mining technology.Therefore,the author uses the related data he has sorted in his hand as the dataset and attempts to use TF-IDF(term frequency-inverse document frequency)and BERT(Bidirectional Encoder Representations from Transformers)methods combined with relevant terminology to achieve knowledge clustering based on title similarity,providing a reference for comprehensive knowledge clustering in the later period.Experimental results show that the TF-IDF method tends to cluster based on title keywords,while the BERT method tends to cluster based on title semantics.
关键词
新中国陶瓷/聚类/BERT/TF-IDF/景德镇
Key words
New China Ceramics/Clustering/BERT/TF-IDF/Jingdezhen
引用本文复制引用
基金项目
江西省教育厅科学技术研究项目(GJJ2201039)
China Ceramic Development Research Institute of Jingdezhen Ceramic University(2023ZK03)
Jiangxi Social Science Planning Commission Project(2023ZK03)
维普
万方数据