Understanding Fake News Geography and Major Determinants of Global News Coverage of Disasters

Motivation

It was always fascinating for me to know how such a huge amount of data is generated every minute and how this data is handled and further processed for improvement and perfection of human life. Big data and data mining is taking the world by storm through its numerous benefits. It is allowing many of the leading firms to develop some of the cutting-edge technologies providing high-end services to their businesses.

Strategy used by me for searching papers on this topic was the arxiv.org website. This is one of the largest collection of published papers. I tried searching for papers related to big data and then found that author named Haewoon Kwak had done a lot of research in this field. I also had a look to other articles by this author. Finally, I choose this paper after going through many other research papers in domain of ‘Big Data and Data Mining’. One of the main reason for selecting this was the idea behind author’s research which depicts how the world’s largest available event dataset can be used for deciding numerous factors. It has shown how news media from different countries play a major role in global news coverage. Authors also analyzed the structure of disaster’s news geography depending upon occurrence of disasters and their coverage by relevant countries. Finally using regression model, major determinants were found out. Explanatory power of these variables shows how they are responsible for global newscast reportage of disasters.

This is an interesting research and can be further extended for other use-cases for example prediction of disaster depending upon historical disaster data provided by news. Also other data sources can be taken into consideration like data from social media. According to research such sources provides more accurate data as it contains user chatters which are sometimes more quick as compared to any news media.

Goal of research

Nowadays open sources like GDELT contains vast amounts of data containing many significant KPI’s and other parameters. Such datasets can be easily used to derive many useful use-cases.

The main aim of authors behind this research was to divulge the organization of global news coverage and find the major determining factors behind this news geography of disasters. They wanted to discover how important a variable’s unique contribution may be by building a hierarchical multiple regression model.

Infographics

Abstract

This work uses GDELT (Global Data on Events, Location and Tone) for revealing the structure of global news coverage. Important variables in the hierarchical regression model, such as political stability, GDP, population, GNI, damage, etc. are nicely associated with the researches that were done previously on this topic. On the other hand, authors find very much regionalism while detecting the news geography. This generates the need to take into consideration a wide-ranging and complete data for further analyzing and researching on the coverage of news globally.

Keywords

GDELT Project, Big data and data mining, news geography, organization of global news coverage, regionalism, theory of newsworthiness, international news agency, foreign news.

Some of the above keywords are explained below:

News geography and regionalism

News geography is illuminating and creating a general pattern which shows which all countries are present in which others country’s news representation. In international study of news this is one of the major issues which researchers are facing. Authors have associated their verdicts to news geography’s well established frameworks. Generally strong regionalism is found in this geography of news due to news focusing countries and this is a major reason we require comprehensive datasets.

Theory of news worthiness

It is necessary to make a decision on which news should be taken into consideration and which news do not have that much importance. Studies conducted by Galtung and Ruge examines some of the aspects of news worthiness. According to them individual’s psychology plays a major role in determining the newsworthiness of an incident.

After their studies, they came up with certain factors which really decides which news to be considered worthy. Some of these factors are meaningfulness, intensity, ambiguity, identity of the actors involved, other characteristics of the actor, for how long the event was continued, etc. Several factors amongst these were taken from GDELT dataset for further research. Importance of some of the factors like number of people killed is still disputable. Still this factor is a good one to analyze the outcome especially if other factors are unavailable. It is included by authors in the study.

The GDELT project

GDELT stand for Global data on event, location and tone. It is a newly developed event dataset and at present it contains more than 200 million geolocated events. They are tracking news all over the world in 100 different languages and their sources range from BBC news, Africa News, Agence France Press, Washington Post, google news, etc.

This project collects information from news media and stores it in CAMEO (Conflict and Mediation Event Observations) event coding standard. It is categorized in CAMEO format using the open source TABARI (Textual Analysis by Augmented Replacement Instructions) systems.

Using pattern recognition, TABARI system machine codes the International event data. It is a successor to KEDS (Kanas Event data system) which was developer during late 1990s. As compared to human coding, machine coding provides some advantages and that is why TABARI got recognition. One advantage is quickness. Coding can be done more quickly by machines as compared to humans. Also rules of coding are applied with consistency and do not affect by inter code discrepancies.

Due to such advanced systems, GDELT performs better as compared to other datasets.

It provides two datasets:

• Event dataset: This is a smaller dataset containing event details like actors, countries, etc. coded by CAMEO taxonomy.
• GKG (Global knowledge graph): It is an expanded dataset containing very small details about person, location, organization and more than 230 emotions and themes from all the reported news.

Authors use GKG as it offers various fields like type of disaster, number of victims, number of news articles representing the disaster, etc. which is useful for study.

News Geography of disasters

Authors concentrated on news geography where the extent to which countries are shown in other country’s news is depicted. They tried to show global representation of a particular event in a country. For this, initially they divided the world into seven different regions based upon their geographical proximity.

These regions are:

1. East Asia & Pacific,
2. Europe & Central Asia,
3. Latin America & Caribbean,
4. Middle East & North Africa,
5. North America,
6. South Asia,
7. Sub-Saharan Africa

Based upon the visitor’s nationality, they have mapped altogether of 10,009 media from different countries in the world. Then they have a list of news media falling into each specified regions.

For the sake of this calculation authors have defined Attention of a region ri, to the country cj, as the number of disasters occurring in a country which are covered by or reported by the news media falling in that region. This is represented using the notation Nri ? cj.

News geography observed by the region as

Nri = {Nri ? c1 Nri ? c2, Nri ? c3, ……., Nri ? ck},

Where k are the number of countries taken into consideration. Authors used Cartogram to show the news geography seen by each region. It precisely shows the relation between territories of a country as compared to its value.

In this diagram we can clearly see the differences between representations of countries as compared to other countries. Some countries are over represented in the regions they occur. This is because of the reporting of disasters happening in that country by the news media lying in that region for example USA in N North America, India in N South Asia, Saudi Arabia in N Middle East. 

Cite this page