question about importing from "Twitter Search Network"

Apr 16, 2013 at 12:01 AM

I downloaded Twitter data with a search word, but the edge list displays self-loops for re-tweets and mentions. Follower relationships look ok.

For example,

LaLa_Loves tweeted "RT @nytimes: President Obama: "We will find out who did this, and we will hold them accountable." While I feel that the edge list should include Lala_Loves as Vertex 1 and nytimes as Vertex 2, it has Lala_Loves as vertex 1 and again Lala_Loves as vertex 2. The visual, therefore, results in a bunch of isolates with self-loops.

Did I miss any step? Or, is it something that I should manually code based on the tweets found?



Apr 16, 2013 at 2:45 AM
Edited Apr 16, 2013 at 2:46 AM

NodeXL will create an edge with a "Relationship" of "Mentions" only if the tweeter mentioned someone else in the network. Similarly, it will create an edge with a "Relationship" of "Replies-to" only if the tweeter replied to someone else in the network.

People are in the network only if they have tweeted the search term you specified. Another way of saying this is that NodeXL creates vertices only for people who tweeted the search term.

In the example you provided, @nytimes did not tweet the search term you specified, and so @nytimes is not in the network. Instead of discarding the tweet from LaLa_Loves, NodeXL created a self-loop edge with a "Relationship" of "Tweet". When you look at these edges in the graph you might conclude that these people were tweeting only to themselves, but what they really represent are tweets that didn't mention or reply to other people in the network.

-- Tony
Apr 16, 2013 at 5:12 PM
Thanks Tony! That explains very well :)