What are the patterns of image diffusion during the Nepal Earthquake?
by Heather Witte, Kari Santos, Denis Kazakov, Fadhil Suhendi, Zach Lamb

Questions

What language are the tweets in this dataset written in?(contributed by Heather)

What were average retweet ratios on each day?(assumption that as days pass on, people scroll twitter less => less retweets) (contributed by Denis)

What user's tweet got retweeted the most? (contributed by Zach)

How many original tweet in location Kathmandu, Nepal? (contributed by Fadhil)

For retweeted images, does the number of followers of the original tweeter have a direct effect on the number of retweets? (contributed by Kari)

X axis = number of followers
Y axis = retweet count
It doesn't seem that the number of followers of the original tweeter has an effect on the number of retweets.
# Discussion The answers submitted here probably cause more additional questions to be raised than full answers. For instance, what causes a tweet with an image to be retweeted? Is it the "interest" of the image? It doesn't seem to correlate with the number of followers of the original tweeter, but what if you looked at the follower count of the retweeters? If a tweet is retweeted by someone with a large following at any time in the retweet path, perhaps that's what makes it go viral. Do tweets with comments in less popular languages have less of a chance of going viral? What is the typical "lifecycle" of a tweet and are there common patterns? How quickly to tweets leave the location of a disaster? For example, I began wondering how often users tweet in a language other than their user setting language, but found that all tweets in the dataset fit that criteria, since users tweeted in either Nepalese or Hindi, while their user setting was English. Since this finding was uninteresting, I decided to change my question to simply analyze the proportion of languages that users tweeted in. I think that these types of language-based questions would be interesting to apply to the entire set of tweets from the Nepal earthquakes, to see the patterns of diffusion across the entire world, and not just on the set of local tweets. More tweet data over a longer duration and a second round of analysis would prove very interesting.