Internet Geographer

Blog

Posts tagged twitter
The Geography of Twitter

A few months ago, Antonello Romano and I published some maps of Twitter. Those maps showed which parts of the world produced more content than others. However, what they failed to do is account for differences in Internet penetration around the world. 

The above map normalises the Twitter data by internet population data: revealing the parts of the world that are home to internet users who are more likely to publish content on the platform. 

You can see that the differences between places are not slight ones. Internet users in some countries (like Malaysia) are dozens of times more likely to tweet than internet users in places like India or Kenya. 

As in painfully obvious in 2017, information in social media streams can have an outsized influence. Knowledge shared on Twitter can shape how people around understand society, the economy, and politics. But, as we see here, that knowledge has distinct geographies. It is far more likely to be created in some places than others.

Further reading:

Graham, M, S. Hale, and D. Gaffney. 2014. Where in the World are You? Geolocation and Language Identification in Twitter. The Professional Geographer 66(4) 568-578. (pre-publication version here)

Graham, M., De Sabbata, S., Zook, M. 2015. Towards a study of information geographies:(im)mutable augmentations and a mapping of the geographies of information Geo: Geography and Environment.2(1) 88-105. doi:10.1002/geo2.8

Graham, M. 2015. Information Geographies and Geographies of Information New Geographies 7 159-166.

Graham, M., S. Hale & M. Stephens. 2011. Geographies of the World's KnowledgeConvoco! Edition.

infogeog, oiiMark Grahamoii, twitter, map
Mapping Twitter

I've been working with Antonello Romano to update some of our older research into the geography of Twitter

Below you can see some maps from a sample of about 2.5 million tweets collected worldwide over 48 hours in October 2016. These are collected using the Twitter streaming API (at most a 1% sample).

Because of the nature of the data, I wouldn't read too much into any specific differences. But what these maps broadly achieve is that they give us a sense of the digital cores and peripheries of our world.

The two 'world at night' style maps give a sense of which parts of the world light up the internet with content, and which parts are still relatively left in the dark. The choropleth map (the one in which countries are shaded) then gives us a sense of which countries produce the bulk of content. The US (25%) and Brazil (14%) together produce more than a third of the world's content. 

Other parts of the world produce only a tiny amount of content in comparison. In all of Africa combined (2.7% of the world's total), there are is less content produced than in Turkey (4%) or Spain (3%).

We live in a world in which almost half of humanity is connected to the internet. And almost anyone, anywhere, in any position of power is connected. This means that information in social media streams (like Twitter) can have an outsized influence. Knowledge shared on Twitter can shape how people around understand society, the economy, and politics. But, as we see here, that knowledge has distinct geographies. Let's remember that even in 2016, this is anything but a truly global network. 

Further reading:

Graham, M, S. Hale, and D. Gaffney. 2014. Where in the World are You? Geolocation and Language Identification in Twitter. The Professional Geographer 66(4) 568-578. (pre-publication version here)

Graham, M., De Sabbata, S., Zook, M. 2015. Towards a study of information geographies:(im)mutable augmentations and a mapping of the geographies of information Geo: Geography and Environment.2(1) 88-105. doi:10.1002/geo2.8

Graham, M. 2015. Information Geographies and Geographies of Information New Geographies 7 159-166.

Graham, M., S. Hale & M. Stephens. 2011. Geographies of the World's KnowledgeConvoco! Edition.

Hashtags and Haggis: Mapping the Scottish Referendum

Reposted from our work over at Floatingsheep

The past weeks have been quite eventful in Scotland as a monumental election unfolds. Everyone wants to know, which way will the Scots vote? While we here at Floatingsheep certainly don’t have the answer or power to predict the referendum, we thought it might be interesting to see the geographic dimension of how Scots (and the rest of the world) are tweeting about a fundamentally geographic decision [1].

We pulled data from DOLLY from the last month and a half for a number of hashtags and terms that we thought might be helpful in taking the pulse of Twitter discussion around the independence referendum. Most obviously, we collected the hashtags #VoteYes and #YesBecause due to their association with the pro-independence movement, and the hashtag #NoThanks because of its association with anti-pro-independence sentiment [2].


We started by comparing the prevalence of ‘no’ (i.e., pro-union) hashtags versus 'yes’ (i.e., pro-independence) hashtags the global level. In the map below, orange indicates a greater prevalence of 'yes’ tweets and purple indicates that there are more 'no’ tweets. Perhaps the most interesting thing here is that we can see the United Kingdom swing towards a 'yes’ vote, which has, for the most part, appeared to be the underdog in more conventional polling leading up to the referendum. Then again, most of Western Europe, along with Thailand and Australia, also have a general preference for 'yes’ tweets. Oddly enough, the United States is the staunchest defender of the union, based solely on it’s massive preference for 'no’ tweets. Strange for a country that yearly celebrates its breaking away from Mother England

Comparing 'Yes’ vs. 'No’ Tweets at the Global Scale

Looking closer at the UK, we can see that much of Scotland has a roughly equal number of tweets in support of both the 'yes’ and 'no’ positions – reflecting the contentious and hotly-contested nature of this referendum. But the Central Belt in particular – where a lot of actual votes will be coming from, as it is the most densely populated part of the nation – swings heavily towards 'yes’. The English, on the other hand, seem very much inclined towards pro-union or anti-separation tweeting.

Comparing 'Yes’ vs. 'No’ Tweets in the United Kingdom

To take an alternative look at support for the different positions, we mapped the percentage of each of the three hashtags that originates in each of the administrative sub-regions of both Scotland and the UK as a whole. The Highlands and parts of the Central Belt again show up as strong bastions of 'yes’ votes.

Percentage of Referendum-Related Tweets from Different Regions

But seeing as we’re interested in doing more than just mapping distributions, the next question is how are we to put all of this into context? The only proper place to start is, of course, with the Queen. The map below illustrates those places which also tend to have higher-than-normal levels of tweeting about the Queen (in orange) and those places that are tweeting less about the Queen than might usually be expected (in purple), based on a baseline measure of tweeting activity. Sadly, the whole country seems to be ignoring her. Apart from Glasgow, that is. In the interests of not upsetting an 88 year-old lady, we have chosen not to explore these tweets in any more detail.

Tweets referencing “Queen”

Building on this, we also explored the geography of references (using the same method described above) to something inherent in most people’s definitions of Britishness: tea and crumpets

We see an all-around tea-depression; hardly anywhere is particularly pro-tea at the moment, truly a shocking state of affairs. The British are clearly not being their usual selves, and for their sake we’re glad the referendum will be over soon, regardless of the outcome. Scotland, in particular, has average tea counts that are low by historical standards.

Tweets referencing “tea and crumpets”

This analysis would, of course, all be meaningless unless we mapped the geographies of a range of uniquely Scottish phenomena: haggis [3], kilts and Nessie. Still using the same method as above, the map below shows without a shadow of a doubt that Scotland is destined to become it’s own nation.

Tweets referencing “haggis”, “kilts” or “Nessie” 

The Scots are tweeting about these topics at a greater-than-usual rate, while their southern neighbors remain distinctly uninterested. If ever there were an indication that these nations are divided by more than just a line on a map, we see that manifested in the topic of people’s Twitter conversations. In short, the Scottish referendum is not just simply about “yes” or “no” but seemingly touches on much more fundamental questions of ovis-based cuisine, men’s wear and mythological creatures.

So even if the 'no’ votes win out in and the Kingdom remains united, the geographies of haggis related tweeting (along with a few other things) has revealed that these are two very different nations, indeed.

————–
[1] In case you don’t know what Twitter, is we refer you to the Scots Wikipedia page on the subject, which states: “Twitter is an online social networkin service an microbloggin service that enables its uisers tae send an read text-based messages o up tae 140 characters, kent as 'tweets’”.
[2] Perhaps we could have simplified this phrasing, but then we would have lost the chance to type “anti-pro-independence”, which is a lot of fun. Anti-pro-independence. Anti-pro-independence.
[3] Normally the Floatingsheep collective avoids conversation about sheep heart, liver, and lungs that are boiled in a sheep stomach. But we made an exception this time.
New paper: Where in the World are You? Geolocation and Language Identification in Twitter

Scott Hale, Devin Gaffney and I have a forthcoming paper in The Professional Geographer on the geography of Twitter.

The abstract is below, and you can download the pre-publication version from the link at the end of this post.

Abstract

The movements of ideas and content between locations and languages are unquestionably crucial concerns to researchers of the information age, and Twitter has emerged as a central, global platform on which hundreds of millions of people share knowledge and information. A variety of research has attempted to harvest locational and linguistic metadata from tweets in order to understand important questions related to the 300 million tweets that flow through the platform each day. However, much of this work is carried out with only limited understandings of how best to work with the spatial and linguistic contexts in which the information was produced. Furthermore, standard, well-accepted practices have yet to emerge. As such, this paper studies the reliability of key methods used to determine language and location of content in Twitter. It compares three automated language identification algorithms to Twitter’s user language interface setting and to a human coding of languages, and identifies common sources of disagreement. The paper also demonstrates that in many cases user-entered profile locations differ from the physical locations users are actually tweeting from. As such, these open-ended, and user-generated, profile locations cannot be used as useful proxies for the physical locations from which information is published to Twitter.

Full Article

Graham, M, S. Hale, and D. Gaffney. 2013. Where in the World are You? Geolocation and Language Identification in Twitter. The Professional Geographer (in press).