Thursday, March 15, 2012

Retweet Analysis for Two Situation Comedies

The Gephi Retweet Monitor Plugin provides a very convenient way to collect real-time data from Twitter.com. The Retweet Monitor Plugin uses Twitter Streaming API to collect large amount of data.

Approach and Definition of the Graphs
Running the program for consecutive two days, we are able to collect 373 retweets for TwoandAHalfMen and 724 retweets for HIMYM. In the two graphs, the definitions for nodes are Twitter users and the edges are the connections between tweets and retweets.
The two graphs are directed graphs, meaning the edges have directions. For example, if node A retweets node B, there will be an arrow pointing from A to B, making a directed edge. Therefore, node A has one out-degree and user B has one in-degree.
The sizes of the nodes are defined as the level of In-Degree. For example, if node A has more in-degree than node B, node A will have a larger node size.
We analyze the two graphs using “Force Atlas” layout. Both of them display round circle pattern, with the most important nodes lying in the center.


Basic Attributes Analysis
The two graphs have the similar features in attributes with only small differences, Ref Fig 3. This is because both of them are retweet graphs and the observations for the retweet graphs are:

  • Not-well-connected among nodes
  • Large amount of communities within the network

The following table compares the two graphs in varies attributes. It is easy to identify that two graphs have low Degree Centrality of 0.713 for HIMYM and 0.607 for TwoandAHalfMen, which means most of the nodes in two graphs do not have much links to other nodes. The same result can be obtained from Network Diameter for the two graphs, 3 for HIMYM and 2 for TwoandAHalfMen. This indicates within minimum of 3 steps, one node in HIMYM can reach any other nodes within the network, which also proves the network is not well connected.


Parameter
HIMYM
TwoandAHalfMen
Number of Nodes
1448
746
Number of Edges
1032
453
Average Degree
0.713
0.607
Network Diameter
3
2
Graph Density
0
0.001
Modularity
0.919
0.993
Weakly Connected Components
424
293
Avg. Clustering Coefficient
0.002
0
Avg. Path
1.053
1.002
Number of Communities
426
293
Number of shortest paths
1087
454


The two graphs have 424 and 293 connected communities for HIMYM and TwoandAHalfMen. Compared to the total of 1448 nodes in HIMYM, each community only has 3 nodes on average. Similarly, for TwoandAHalfMen, each community only has 2 nodes on average.


Key Findings of Two Graphs
The followings are the similar patterns of the two graphs:
Large amount of isolated clusters
The official sitcom account may not be popular one
Celebrity account received more retweets
Accounts with funny names received more retweets
Both the HIMYM and the TwoandAHalfMen graphs have more scattered nodes than the connected nodes. This is because, in most cases, users can only see the tweets what they follow. Therefore, users can only retweet what they the get from the following accounts.
Official sitcom account and celebrity accounts tend to have more retweets. This is because most nodes do not have much influence in the twitter network. From observations, nodes that had most retweets are from most influential accounts, e.g. @TrechosSerie in TwoandAHaflMen graph and @alecattelan in HIMYM graph.
Besides influential accounts, users in twitter tend to also follow interesting accounts.  Interesting people will have funny username and also tend to have funny tweets. Users tend to follow humorous people and retweet a lot. The typical example is @waitforBarney in HIMYM.


Interesting Observations

  • WGN beats CBS on How I Met Your Mother

How I Met Your Mother is originally shown in CBS. WGN later bought the license to broadcast this sitcom. However, according to our findings in HIMYM graph, @MeetatMacLarens (WGN account) has more retweets than @HIMYM_CBS (CBS account). Ref Fig 4
People are addicted to sitcoms that they want to watch it every day, even though it is an old episode. WGN had a very effective strategy to attract audience from CBS to its own station by paying licensing fees. From a long term perspective, WGN will gain competitive advantage over CBS in terms of licensing expense.

  • Two and A Half Men is more popular in non-US audience

The top three accounts with retweets are from abroad. The tweets and retweets are written in a foreign language. The retweets from @LauraBozzoo was written in Spanish and @MySerieFrases retweets were written in Portuguese.
For CBS, it is a great opportunity to broaden the international market by launching more marketing campaigns on Two and a Half Men in areas in Europe and Asia.


Appendix
Graph for How I Met Your Mother retweets collected for 2 days



Graph for Two and A Half Men retweets collected for 2 days