overview
Twitter reacts to #AylanKurdi, the Syrian toddler refugee who drowned
Twitter reacts to #AylanKurdi, the Syrian toddler refugee who drowned investigates a story based on the following claim:
RIP #AylanKurdi - this should not be happening :-( http://t.co/fOkcRqUnh0 - @K9_Tracks
Summary
investigating

keywords
aylankurdi aylan kurdi

created by
admin

tags

refugees

not showcased

h-index
103

spread
high

level of skepticism
undisputed

collection date
10:04 AM - 3 Sep 2015

tweet count
71,509 (out of 78,771 total)

user count
54,932 (out of 58,793 total)

optional kw threshold
0

contains all required?
false

search count
50,000

date restrictions
None to None

search terms
aylankurdi, aylan kurdi

search term counts
aylankurdi (31134 tweets), aylan kurdi (48937 tweets)

Events & Memes: Twitter reacts to #AylanKurdi, the Syrian toddler refugee who drowned

This work was funded by the National Science Foundation (Grant No. 1117693) and Wellesley College


Welcome to TwitterTrails, a system to investigate the spread and validity of stories on Twitter. TwitterTrails gathers data about news stories, rumors, events, and memes on Twitter, to present in useful and meaningful visualizations that can help users answer questions about how the story spread. Scroll down for the visualizations, or click on “overview” on the top left of the page to view data about this story. For more information about the specific visualizations or the TwitterTrails system, please read our blog or follow us on Twitter .

This page, created automatically by TwitterTrails at 10:04 AM on 3 Sep 2015, investigates a story based on the following tweet:

RIP #AylanKurdi - this should not be happening :-( http://t.co/fOkcRqUnh0 -

Data collected were tweets posted for about the week prior to the start of the investigation. During that time, propagation of this story was high, and the information presented was generally undisputed.


Who broke the story and when?

The Propagation Graph highlights the tweets which were influential in “breaking” the story on Twitter, and highlights independent content creators.

Each circle on the Propagation graph represents a tweet, and hovering over or clicking on the circle will display the tweet to the right of the graph.

Tweets are plotted on x-axis of the graph based on the time they were posted, and on the y-axis by the number of retweets they have received (at the time of data collection).

Circles are sized based on the number of followers the user who posted the tweet has. Circles are drawn by default as gray. Circles with other colors represent tweets with nearly identical texts.

Additionally, circles with a bright blue border indicate tweets written by verified accounts.

Propagation Graph

Who and when the story originated? Is it still spreading?

The Time Series shows the activity over time of relevant data collected.

Time is on the x-axis and the number of tweets generated is on the y-axis. Each point represents a ten minute time span.

Selecting a point on the time series will display on the right a list of the tweets at that time span. These tweets are sorted by the number of retweets they have received (highest on top), and can be re-sorted using the drop down menus. If there are more than 50 tweets in the time span, links to navigate the tweets 50 at a time are provided.

You can zoom in on the graph by clicking and dragging your mouse over a period of time.

Clicking on Manage Series on the bottom right of the display will open a panel which you can use to add new time series to the graph by checking the box on the left.

The Search field takes a search term and will display all tweets contain (the exact) search term when you check the box on the left of Search.

The shape of the Time Series indicates how the story was spreading and whether it was still spreading at the time of the investigation.

Time Series of Relevant Tweets
Manage Series
Count activity every minutes

Relevant
Negation
Non-negation
Retweets
Original
Estimation
Spam
Search:
By User:


Who are the main actors of the investigation?

The co-retweeted network shows the clusters, communities that participate in this investigation, and highlights influential accounts in the retweet network. It is generated by connecting and clustering accounts based on mutual retweeting by other users. (That is, if User A and User B in the co-retweeted network are connected by an edge, it means at least one other user (part of the “audience”) has retweeted both User A and User B. The more members of the audience are retweeting both User A and User B, the stronger the edge among them, and the closer they appear in the cluster.)

Clusters look like clouds and are forming automatically (based on the force-directed algorithm) and indicate the strongest agreement regarding the topic being investigated.

Communities are often parts of the cluster clouds and are colored automatically (based on the Louvain algorithm) and indicate similarity between users in the community: they have stronger connections within their community than outside of it. There could be several communities within a cluster.

Hovering over a point will display the name of the influential account it represents, and clicking on it will bring up information about that account on the right of the graph. The user information also contains the tweets written/retweeted by that user in the dataset.

Partition Graph by:

user language

verified vs. unverified

negation (scaled)

reset
Co-Retweeted Network

How are community members describing themselves?

There are a total of 49 communities of similar users in the Co-Retweeted Network. The largest community has 593 users in it, and the smallest has 2. Nodes in the co-retweeted graph are colored based on their community.

Each community is also represented by a word cloud in the colored rectangle below. The more often community members use a word in their profile, the larger that word appears in the word cloud.

To view aggregation statistics about any of the communities, you can either click on a node in the graph, or select a community from the panel below.

Co-Retweeted Network Statistics

What pictures were used in the tweets?

Research has shown that a picture is a powerful way to promote a message because it has strong emotional impact on people. This section shows the most retweeted images are displayed in this investigation. Hover your cursor over on an image to see the tweet in which that image was posted.

In Pictures