GDELT数据网络化挖掘与国际关系分析

doi:10.12082/dqxxkx.2019.180674

摘要/Abstract

摘要：

21世纪以来的国际关系错综复杂、瞬息万变,给世界的经济、安全、外交等带来了深刻变化。这些变化对中国的内外政策产生了重大影响。全面及时地分析国际关系及其变化特征,对于中国的经济和外交发展规划具有重要参考价值。国际关系研究具有复杂性、及时性、时空性等特点,迫切需要时空大数据分析技术为其提供新的思路和技术手段。大众媒体如报纸、广播等记录着世界上发生的各种各样的事件,蕴含着丰富的信息,相对于记录个人活动的社交媒体数据,其更加适合于对人类社会进行大规模和长时间的分析。GDELT是一个免费开放的新闻数据库,它实时监测世界上印刷、广播、网络媒体中的新闻,对其进行文本分析并提取出人物、地点、组织和事件等关键信息。本文利用复杂网络的理论和方法对GDELT进行网络化挖掘并进一步分析国家关系。首先利用该数据构建国家交互网络,然后通过网络特征统计分析国家之间的交互关系,最后探测国家冲突事件交互网络的时序变化。研究发现：① 国家交互网络具有无标度特性,网络连接在整体和局部上都呈现出不均匀性,少数国家与其他国家有大量交互,大多数国家与其他国家的交互很少;一个国家与少数国家有大量交互,而与大多数国家的交互很少。② 国家冲突事件交互网络的突然变化往往对应一些重大事件。本文的研究可以为大数据时代的国际关系探索提供一个新的视角,同时也为新闻媒体数据的分析提供参考。

关键词: GDELT, 空间交互网络, 时空大数据分析, 国际关系, 复杂网络, 无标度分布

Abstract:

The international relations are intricate and ever-changing since the 21^st century, and have brought profound changes to the world's economy, security, and diplomacy. These changes have had a major impact on China's internal and external policies. A comprehensive and timely analysis of international relations and its changing characteristics has important reference value for China's economic and diplomatic development planning. The analysis of international relations has spatio-temporal characteristics, and it needs real-time processing. Thus, it needs to introduce the methods of spatio-temporal big data analysis to analyze international relations. Traditional mass media such as news, radio, etc. record all kinds of events happening in the world. It contains a wealth of information. Compared with social media data recording personal activities, it is more suitable for large-scale and long-term analysis of human society. The Global Database of Events Language, and Tone (GDELT) is a free and open news database which monitors news from print, broadcast, and online media in the world, analyzes the texts and extracts the key information such as people, place, organization, and event. This paper researches the network characteristics of GDELT based on theory of complex network and further analyze the relations between countries. Firstly, this paper constructs national interaction networks using GDELT, then analyze the interaction relationship between countries through network characteristic statistics, and finally detect the time series changes of the national conflict event interaction network. The results show that: (1)The National interaction network has scale-free characteristics, the interaction between countries is unevenly distributed from a global and local perspective. Very few countries have lots of interactions while most countries have very few interactions, and one country has lots of interactions with a few countries while a few interactions with most countries. (2) Sudden changes in the national interaction network of conflict events often indicates some significant national conflict events. This paper can provide a new perspective for the exploration of international relations and a reference for the analysis of news media in the era of big data.

Key words: GDELT, spatial interaction network, spatio-temporal big data analysis, international relations, complex network, scale-free distribution

秦昆, 罗萍, 姚博睿. GDELT数据网络化挖掘与国际关系分析[J]. 地球信息科学学报, 2019, 21(1): 14-24.DOI:10.12082/dqxxkx.2019.180674

Kun QIN, Ping LUO, Borui YAO. Networked Mining of GDELT and International Relations Analysis[J]. Journal of Geo-information Science, 2019, 21(1): 14-24.DOI:10.12082/dqxxkx.2019.180674

图/表 11

表1

表2

图1

表3

图2

图3

图4

图5

图6

图7

图8

参考文献 37

[1]	Lazer D, Pentland A, Adamic L, et al.Computational social science[J]. Science,2014,323(1):721-723.
[2]	Kang C, Qin K.Understanding operation behaviors of taxicabs in cities by matrix factorization[J]. Computers Environment & Urban Systems, 2016,60:79-88. doi: 10.1016/j.compenvurbsys.2016.08.002
[3]	Eagle N, Pentland A S, Lazer D.Inferring friendship network structure by using mobile phone data[J]. Proceeding the National Academy of Sciences of the United States of America, 2009,106(36):15274-15278. doi: 10.1073/pnas.0900282106 pmid: 19706491
[4]	Mazzitello K I, Candia J, Dossetti V.Effects of mass media and cultural drift in a model for social influence[J]. International Journal of Modern Physics C, 2007,18(9):1475-1482. doi: 10.1142/S0129183107011492
[5]	The GDELT Project [EB/OL] [2018-12-27].
[6]	Leetaru K, Schrodt P A.Gdelt: Global data on events, location, and tone, 1979-2012[C]//ISA annual convention. 2013,2(4):1-49.
[7]	Kwak H, An J.A first look at global news coverage of disasters by using the gdelt dataset[C]//International Conference on Social Informatics, Cham: Springer, 2014:300-308.
[8]	Su Y, Lan Z, Lin Y R, et al.Tracking public response and relief efforts following the 2015 Nepal earthquake[C]// IEEE, International Conference on Collaboration and Internet Computing, IEEE, 2016:495-499.
[9]	Degtyarev D, Badrutdinova K, Stepanova A.Interconnections among the United States, Russia and China: Does Kissinger’s American leadership formula apply?[J]. International Organizations Research Journal, 2017,12(1):81-109. doi: 10.17323/1996-7845
[10]	方鹏. 基于Django的海量媒体数据分析平台的设计与实现[D].南宁:广西大学,2017.
	[ Fang P.The design and implementation of massive data analysis platform based on Django[D]. Nan Ning: Guangxi University, 2017. ]
[11]	Bi S, Gao J, Wang Y, et al.A contrast of the degree of activity among the three major powers, USA, China, and Russia: Insights from media reports[C]// International Conference on Behavioral, Economic and Socio-Cultural Computing. IEEE, 2015:38-42.
[12]	Sagi D J B, Labeaga J M. Using GDELT data to evaluate the confidence on the Spanish government energy policy[J]. International Journal of Interactive Multimedia and Artificial Intelligence, 2016,3(6):38-43. doi: 10.9781/ijimai.2016.366
[13]	龚为纲,朱萌.社会情绪的结构性分布特征及其逻辑——基于互联网大数据GDELT的分析[J].政治学研究,2018(4):90-102.
	[ Gong W G, Zhu M.The hierarchical distribution characteristics of social emotions and the explanation: Analysis based on enternet big data set GDELT[J]. CASS Journal of Political Science, 2018(4):90-102. ]
[14]	Alikhani E.Computational social analysis: Social unrest prediction using textual analysis of news[M]. New York: State University of New York at Binghamton, 2014.
[15]	Yonamine J E. Predicting future levels of violence in Afghanistan districts using gdelt [EB/OL][2018-12-27]. .
[16]	Qiao F, Li P, Zhang X, et al.Predicting social unrest events with hidden markov models using GDELT[J]. Discrete Dynamics in Nature and Society, 2017,2017:1-13.
[17]	Keneshloo Y, Cadena J, Korkmaz G, et al.Detecting and forecasting domestic political crises: A graph-based approach[C]// ACM Conference, ACM,2014:192-196.
[18]	Qiao F, Wang H.Computational approach to detecting and predicting occupy protest events[C]// International Conference on Identification, Information, and Knowledge in the Internet of Things, IEEE, 2015:94-97.
[19]	Elshendy M, Fronzetti Colladon A.Big data analysis of economic news: Hints to forecast macroeconomic indicators[J]. International Journal of Engineering Business Management, 2017,9:1-12. doi: 10.1504/IJBE.2017.084703
[20]	Phua C, Feng Y, Ji J, et al.Visual and predictive analytics on Singapore news: Experiments on GDELT, Wikipedia, and^STI[EB/OL][2018-12-27]. .
[21]	Elshendy M, Colladon A F, Battistoni E, et al.Using four different online media sources to forecast the crude oil price[J]. Journal of Information Science, 2018,44(3):408-421. doi: 10.1177/0165551517698298
[22]	Sharma K, Sehgal G, Gupta B, et al.A complex network analysis of ethnic conflicts and human rights violations[J]. Scientific Reports, 2017,7(1):8283. doi: 10.1038/s41598-017-09101-8 pmid: 5557885
[23]	Yuan Y, Liu Y, Wei G.Exploring inter-country connection in mass media: A case study of China[J]. Computers Environment & Urban Systems, 2017,62:86-96.
[24]	Yuan Y.Modeling inter-country connection from geotagged news reports: A time-series analysis[C]// International Conference on Data Mining and Big Data. Springer, Cham, 2017:183-190.
[25]	汪小帆. 复杂网络理论及其应用[M].北京:清华大学出版社,2006.
	[ Wang X F.Complex network theory and applications[M]. Beijing: Tsinghua University Press, 2006. ]
[26]	Boccaletti S, Latora V, Moreno Y, et al.Complex networks: Structure and dynamics[J]. Physics Reports, 2006,424(4):175-308. doi: 10.1016/j.physrep.2005.10.009
[27]	秦昆,周勍,徐源泉,等.城市交通热点区域的空间交互网络分析[J].地理科学进展,2017,36(9):1149-1157. doi: 10.18306/dlkxjz.2017.09.011
	[ Qin K, Zhou Q, Xu Y Q, et al.Spatial interaction network analysis of urban traffic hotspots[J]. Progress in Geography, 2017,36(9):1149-1157. ] doi: 10.18306/dlkxjz.2017.09.011
[28]	Dueñas M, Fagiolo G.Modeling the international trade network: A gravity approach[J]. Journal of Economic Interaction & Coordination, 2013,8(1):155-178. doi: 10.1007/s11403-013-0108-y
[29]	Davis K F, D'Odorico P, Laio F, et al. Global spatiotemporal patterns in human migration: A complex network perspective[J]. Plos One, 2013,8(1):e53723. doi: 10.1371/journal.pone.0053723 pmid: 3553122
[30]	All GDELT Event Files[EB/OL][2018-12-27]. .
[31]	All gdelt GKG Files[EB/OL][2018-12-27]. .
[32]	Gerner D J, Schrodt P A, Yilmaz O, et al.The creation of CAMEO (Conflict and Mediation Event Observations): An event data framework for a post cold war world[C]//annual meeting of the American Political Science Association, 2002:29.
[33]	Barabási A L, Albert R.Emergence of scaling in random networks[J]. science, 1999,286(5439):509-512. doi: 10.1126/science.286.5439.509
[34]	Zipf G K.Selected studies of the principle of relative frequency in language[J]. Language, 1933,9(1):89-92. doi: 10.4159/harvard.9780674434929
[35]	Arnold B C.Pareto and generalized pareto distributions[M]. New York: Springer, 2008.
[36]	Clauset A, Shalizi C R, Newman M E J. Power-law distributions in empirical data[J]. SIAM review, 2009,51(4):661-703. doi: 10.1137/070710111
[37]	唐莲,王大辉.关于Zipf-Mandelbrot律中参数ρ的一种解释[J].北京师范大学学报(自然科学版),2011,47(1):97-100.
	[ Tang L, Wang D H.A explanation of shift parameter ρ in Zipf-Mandelbrot law[J]. Journal of Beijing Normal University (Natural Science), 2011,47(1):97-100. ]

Actor1Geo_CountryCode	Actor2Geo_CountryCode
CH	US
SY	RS
US	RS

数目	位置
2	1#China#CH#CH#35#105#CH; 1#UnitedStates#US#US#39.828175#-98.5795#US; ;1#Canada#CA#CA#60#-96#CA
4	1#United States#US#US#39.828175#-98.5795#US; 1#Russia#RS#RS#60#100#RS
5	1#United States#US#US#34.04#-118.15#US; 1#Canada#CA#CA#60#-96#CA

数据源	时间	N	M	k	D	C	L	A
事件库	1天	213	2420	22.723	0.107	0.514	2.080	-0.200
	1周	233	5385	46.223	0.199	0.613	1.877	-0.201
	1月	246	8861	72.041	0.294	0.678	1.732	-0.194
全球知识图	1天	242	17 032	140.760	0.584	0.813	1.424	-0.154
	1周	254	25 023	197.031	0.779	0.923	1.222	-0.111
	1月	257	27 960	217.588	0.850	0.941	1.150	-0.059