NETWORK CENTRALITY IN SUB-NATIONAL AREAS OF INTEREST USING GDELT - - PowerPoint PPT Presentation

▶
network centrality in sub national areas of interest
SMART_READER_LITE
LIVE PREVIEW

NETWORK CENTRALITY IN SUB-NATIONAL AREAS OF INTEREST USING GDELT - - PowerPoint PPT Presentation

Aug 22, 2022 386 likes •588 views

NETWORK CENTRALITY IN SUB-NATIONAL AREAS OF INTEREST USING GDELT DATA 11 July 2014 CDT Caitlin Rowe The Power of Partnership from Vision to Reality Introduction Caitlin Rowe Senior at United States Military Academy Studying

slide-1
SLIDE 1

The Power of Partnership – from Vision to Reality

NETWORK CENTRALITY IN SUB-NATIONAL AREAS OF INTEREST USING GDELT DATA

11 July 2014 CDT Caitlin Rowe

slide-2
SLIDE 2

Introduction

  • Caitlin Rowe
  • Senior at United States Military

Academy

  • Studying Systems Engineering

Senior Project/Capstone with Data- Tactics for AY15

  • Goals for 3 week summer

internship

  • Make connections for help during

academic year

  • Learn how to use Shiny
  • Explore GKG data for Dhaka, Bangladesh
  • Explore the utility of subnational

analysis

2

slide-3
SLIDE 3

Agenda

  • Data Analytics in the Military
  • Familiarization with GKG
  • Network Centrality Analysis for Military Application
  • GDELT Analysis Service
  • Shiny and Gephi Results
  • Communicating Conclusions to Military Leadership
  • Future Areas for Improvement

3

slide-4
SLIDE 4

Data Analytics in a Military Context

  • Technological advances
  • Vast amounts of data
  • Sensors
  • Social Media Sites
  • HUMINT
  • Used to increase
  • Efficiency
  • Accuracy

4

slide-5
SLIDE 5

Data Analytics in a Military Context

5

“Simply put, big data is now a fixture

  • n the battlefield and across the

global security landscape.” -Forbes

http://www.forbes.com/sites/techonomy/2012/03/12/military-intelligence-redefined-big-data-in-the-battlefield/

slide-6
SLIDE 6

My Research

  • Guidance from project

advisor

  • GDELT
  • Global Knowledge Graph
  • Dhaka, Bangladesh

6

  • Purpose for Research
  • Assess GKG at subnational

level

  • Determine influential

individuals and

  • rganizations
slide-7
SLIDE 7

Familiarization with Dataset

  • Global Knowledge Graph
  • Date, Event, People,

Location, Tone, and Article(url)

7

  • Collected from web, print, and

broadcast news sources

slide-8
SLIDE 8

Network Centrality for Military Application

8

  • Identify leadership and key players in given area
  • Friendly and Enemy
  • Which centrality measure is the most important in a

military operation?

  • Degree
  • Closeness
  • Betweenness
slide-9
SLIDE 9

GDELT Analysis Service

9

GDELT .org Analysis Service Email containing analysis

ex.filtered data, graph, map

Email address Date range Key words

slide-10
SLIDE 10

GDELT Analysis Service

  • GKG Network
  • Heat map
  • Exporter
  • Geographic Network
  • Word Cloud

10

slide-11
SLIDE 11

GDELT Analysis Service- GKG Network

11

slide-12
SLIDE 12

Important Nodes

12

slide-13
SLIDE 13

Shiny

13

Filter using R and create a network (igraph) All GKG raw data (downloaded) Create Shiny App that displays network, nodes, and centrality measures

slide-14
SLIDE 14

Current Static Network

14

slide-15
SLIDE 15

Gephi

15

Filter using R and create a network (igraph) All GKG raw data (downloaded) Export network as “.graphml” file Open network in Gephi and explore using built in tools

slide-16
SLIDE 16

Gephi Network Example

16

slide-17
SLIDE 17

Gephi Network Example

17

http://webatlas.fr/tempshare/ForceAtlas2_Paper.pdf

slide-18
SLIDE 18

Conclusions

  • Efficiency
  • The GKG Analysis Tools, Gephi, and Shiny all

produce network visualizations that display network centrality. These results can aid military leaders’ decisions of who to target or exploit based on specific mission information.

  • Accuracy
  • All of the network information is gathered from

news sources, so connections between individuals that are clandestine or informal might not be included.

18

slide-19
SLIDE 19

Future Areas of Improvement

  • Shiny App
  • Change city, date range
  • Separate by community
  • Gephi
  • More seamless transition from R
  • Overall
  • Remove newspaper editors

19