中文文献关键词分布特性研究

KEYWORD DISTRIBUTION CHARACTERISTICS OF CHINESE LITERATURE

在中文引文网络中,研究节点之间引用关系特性的成果较多,但是,对引文网络中的关键词研究却非常少见。关键词是论文的代表词语,可通过关键词大致了解论文所研究的重点和领域。因此,对于挖掘关键词的分布特性的研究尤为重要。从中国科学期刊爬取人工智能、生物和财经三个领域的关键词各约2500个;从百度学术、知网和Bing学术搜索引擎中爬取每个关键词的结果数目,对于百度学术搜索引擎中,另外爬取2016年、2017年和2018年等每年的结果数,并爬取每个关键词下的三个相关知名学者;基于以上数据,引入Zipf定律构建结果数与排名之间的关系模型(包括近三年的结果数与排名之间的关系模型);引入超网络模型,构建关键词与相关知名学者之间的超网络模型。基于以上两个模型,分析得出了关键词分布的几个有趣的相关结论。

In the Chinese citation networks,there are many achievements on studying the characteristics of citation relations between nodes.However,it is rare to research the keywords in citation networks.As the representative words of the paper,the keywords can be used to understand the key point and field of the paper.Therefore,it is particularly important to study the distribution characteristics of keywords.In order to realize this research,we selected about 2500 keywords in the fields of artificial intelligence,biology and finance from Chinese science journals,and collected the number of results of each keyword from Baidu Academic,CNKI and Bing academic search engines.For Baidu academic search engine,we searched the annual results of 2016,2017 and 2018 and three relevant well-known scholars under each keyword.Based on the above data,Zipf’s law was introduced to build the relationship model between the number of results and ranking(including the relationship model between the number of results and ranking in the recent three years).The hyper-network model was adopted to construct the hyper-network model between keywords and relevant well-known scholars.According to the above models,we obtain several interesting conclusions about keyword distribution.