摘要 近年来,围绕金融网络进行的构建与分析研究受到学界的广泛关注。现有研究视角主要集中在利用金融市场内的交易数据,构建网络并开展分析;然而,在大数据时代,如何利用包括新闻文本在内的非金融市场另类数据开展研究,仍有较大探索空间。采用文本挖掘中的共现分析思路,提出一套针对中文语境基于事件关联挖掘的金融网络构建方法,并在此网络基础上分别进行拓扑结构分析、金融事件路径传递分析和事件-实体关联映射分析。实验结果表明,基于金融网络构建方法,不仅能帮助理解金融市场演变路径,而且可以在一些情境下更好地识别金融市场参与实体间的关联关系。 In recent years,researches on the construction and analysis of financial network have been widely concerned.The existing researches mainly focus on the use of transaction data in the financial market.However,in the era of big data,how to use non-financial market data,such as news texts,to carry out researches still be a large space for exploration.based on the idea of co-occurrence analysis,this paper proposes a method of constructing financial network based on event association mining in Chinese context.based on the constructed network,we carry out the analysis of topological structure,financial event path transitivity,and event entity association.The experimental results show that our proposed approach based on financial networks can not only help to understand the evolution of financial markets,but also better realize the relationship between the entities involved in financial markets in some situations.
出处 《中国科学院大学学报》 CSCD 北大核心 2021年第2期270-279,共10页 Journal of University of Chinese Academy of Sciences
基金 国家自然科学基金(71450009) 北京市社会科学基金(19YJC036) 北京市教育委员会科技一般项目(KM20190038002) 首都经济贸易大学2019年度科研基金项目(ky2019001132)资助。
关键词 金融网络 共现分析 关联挖掘 实体关联 复杂网络 financial network co-occurrence analysis correlation extraction entity correlation complex network
