面向股票的财经新闻关联度研究——基于新闻价值量化理论  

Stock-Oriented Measurement of Financial News Correlation:Based on Theory of Quantifying News Value

作　　者：石静[1,2] 张斌 陈烨[1,2] SHI Jing;ZHANG Bin;CHEN Ye(Laboratory for Data Intelligence and Cross-Innovation of Nanjing University,Nanjing 210023;School of Information Management,Nanjing University,Nanjing 210023)

机构地区：[1]南京大学数据智能与交叉创新实验室,南京210023 [2]南京大学信息管理学院,南京210023

出　　处：《科技情报研究》2023年第3期36-48,共13页Scientific Information Research

基　　金：江苏省“双创博士”(编号:JSSCBS20210044);国家自科基金面上项目“基于注意力机制的学术信息动态推荐研究”(编号:72074109)。

摘　　要：[目的/意义]为了从海量财经新闻中快速、准确识别与特定股票相关的重要信息,充分挖掘其潜在价值,文章开展面向股票的财经新闻关联度研究。[方法/过程]利用自然语言处理与机器学习方法实现新闻全文本分析,细化到词语粒度挖掘并量化股票-财经新闻关联关系。在此基础上,基于新闻价值量化理论,建立“面向股票的财经新闻关联度测度体系”。[结果/结论]构建了融入新闻价值评价的股票-财经新闻关联度测度体系,实现了个性化、自动化的股票-财经新闻关联程度测量;进一步分析了各个维度指标对关联度测度的影响。[Purpose/significance]To identify crucial financial news information and tap its potential value related to specific stocks more quickly and accurately,we conduct financial news correlation measurement research in terms of stocks.[Method/process]Natural Language Processing and Machine Learning are used to measure the correlation by text analysis in the word's dimension.Then,the theory of quantifying news is applied to construct a stock-oriented evaluation index system for financial news correlation.[Result/conclusion]This paper realizes a personalized and automatic measurement of news correlation with the index system.Further,the influence of each index is also be analyzed.

关 键 词：新闻评估 文本挖掘 股票市场 语义关联 

分 类 号：G353[文化科学—情报学]