Twitter Timeline
- "Charles Darwin's evolution experiment on Ascension isle" - link1 day ago
- Charlie spent his first day at day care today. Was minorly traumatic (for us). Gonna pick him up soon.1 day ago
- @kruel74 @imTheJessicat Thanks much appreciated :) Hope you have a good day too!1 day ago
- Facebook not very good at privacy - again... "Non-Authorized Check-Ins by Friends are Visible" link1 day ago
- @fourfeetnineaud Absolutely - he should have a personal Fail blog :/4 days ago
- Seems like Ibrahim Ali may have crossed a line - wanting to dictate syariah law? link4 days ago
- "Who’s in the Blogosphere?" - Technorati study visualised link4 days ago
- Paris Hilton arrested link5 days ago
|
Other sites
Tags
academic advertising all-blogs ambp anthropology authenticity autobiography belgium blogging blog history blog house blogmeets blogosphere blogs bum 2007 burma camwhoring comments competitions contest culture Current Affairs dogs environment ethics facebook fieldnotes field work film food fun future gardening geekzone google history human rights internet iran israel laos law Malaysia media methodology middle east Miscellaneous monetisation MSM myblogs2009 nab nature nuffnang open source parenting philosophy plastic bags poems politics pps practices random recycle religion research review Sarawak singapore sna sns social media survey technology telephony terrorism Travel twitter usa vista war weaver ants
Quicksearch
Calendar
|
September '10 | |||||
| Mo | Tu | We | Th | Fr | Sa | Su |
| 1 | 2 | 3 | 4 | 5 | ||
| 6 | 7 | 8 | 9 | 10 | 11 | 12 |
| 13 | 14 | 15 | 16 | 17 | 18 | 19 |
| 20 | 21 | 22 | 23 | 24 | 25 | 26 |
| 27 | 28 | 29 | 30 | |||
Blogging Bloggers
As Suanie Sees It
Booey
Changes
Cheeserland.com
*fourfeetnine*
joshuaongys
K A K I B A S I K A L . C O M
KennySia.com
kyspeaks.com
LiewCF.com
Life In The Mist
Make$ Money$
miao~’s comic blog
Nicole Chocoa Heaven
ShaolinTiger
sixthseal.com
SotongZai's Flow In the Ocean
Temptations and Ambitions
The Obnoxious 5xmom
Wattahack? Don't Block My View!!!
YapThomas - The Ultimat3
Booey
Changes
Cheeserland.com
*fourfeetnine*
joshuaongys
K A K I B A S I K A L . C O M
KennySia.com
kyspeaks.com
LiewCF.com
Life In The Mist
Make$ Money$
miao~’s comic blog
Nicole Chocoa Heaven
ShaolinTiger
sixthseal.com
SotongZai's Flow In the Ocean
Temptations and Ambitions
The Obnoxious 5xmom
Wattahack? Don't Block My View!!!
YapThomas - The Ultimat3
Popular Entries
Blogging and Democratization in Malaysia – Forum and book launch
(111784)
Acer Aspire 5580 and Vista Home Premium - the Experience...(83296)
Remixing, blogging and authorship(40490)
Batman haiku(28921)
Blog House / Bangsa Malaysia posts(10811)
Chat boxes and comments(10336)
L - Change the world(10270)
New Acer Notebook and Windows Vista(7121)
Death Race - a review(6966)
Dragonball Evolution Premiere and my Thought Bubble :)(6596)
Last Google Search
Bloggers who hate plastic bags
Tropical Gardening
My other blog 
Categories
Subscribe here
Archives
Powered by
Online PhD
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License
My technical support :) - Andy Ramblings
Social Network Analysis of the Social Media Club - Kuala Lumpur
SMCKL is a group that meets occasionally to explore matters relevant to social media and industry. The most recent one was about social media monitoring tools, and featured three presentations by comScore, Brandtology and JamiQ. They were interesting, but I was surprised that nobody was talking about social network analysis - so I thought I'd do a little demonstration here.
There was much tweeting going on before and after the evening, which was also an occasion for people to meet and network. Using NodeXL, I gathered all the tweets with the hashtag #smckl: in all there were 71 tweeters, and 757 'edges' (i.e. links in the form of 'Followed' relationships, 'Mentions', or 'Replies to'). The following examples only take into account the Followed relationship - i.e. I am only showing a link between tweeters when one follows the other.
A question for social media monitoring has to be: how influential is any particular tweeter? Here I'll look at two ways of visualising that.
Followers
A common measure is how many followers a tweeter has.
In these images, the size of the profile picture is proportionate to the number of followers - the bigger the profile picture, the more followers. Also, the more central the tweeter is, the more ties s/he has with the other tweeters. The person in the middle is the most embedded in the network - with the most ties to other people, directly or indirectly; on the other hand, as you can see, there are some really on the edge - with only a couple of lines attached them to the denser cluster in the middle. They are outliers, less likely to be influential within this group.
The first picture was very dense, so I have filtered out all tweeters with less than 500 followers
and with less than 1000 followers.
Again, a pattern emerges of a denser cluster in the middle with a few outliers. What this suggests is that most people at the SMCKL evening already know each other. But not all: I said above that outliers are less likely to be influential within that group - it's important to note here that the person with the most followers (@victorliew) is an 'outlier'. This suggests that he could be an important 'bridge' for this group to connect to another group. The question would be - who is he? And why are so many people following him?
There was much tweeting going on before and after the evening, which was also an occasion for people to meet and network. Using NodeXL, I gathered all the tweets with the hashtag #smckl: in all there were 71 tweeters, and 757 'edges' (i.e. links in the form of 'Followed' relationships, 'Mentions', or 'Replies to'). The following examples only take into account the Followed relationship - i.e. I am only showing a link between tweeters when one follows the other.
A question for social media monitoring has to be: how influential is any particular tweeter? Here I'll look at two ways of visualising that.
Followers
A common measure is how many followers a tweeter has.
In these images, the size of the profile picture is proportionate to the number of followers - the bigger the profile picture, the more followers. Also, the more central the tweeter is, the more ties s/he has with the other tweeters. The person in the middle is the most embedded in the network - with the most ties to other people, directly or indirectly; on the other hand, as you can see, there are some really on the edge - with only a couple of lines attached them to the denser cluster in the middle. They are outliers, less likely to be influential within this group.
The first picture was very dense, so I have filtered out all tweeters with less than 500 followers
and with less than 1000 followers.
Again, a pattern emerges of a denser cluster in the middle with a few outliers. What this suggests is that most people at the SMCKL evening already know each other. But not all: I said above that outliers are less likely to be influential within that group - it's important to note here that the person with the most followers (@victorliew) is an 'outlier'. This suggests that he could be an important 'bridge' for this group to connect to another group. The question would be - who is he? And why are so many people following him?
The 'Following Differential'
Now, we have all seen 'Followers' like this
Spam tweet accounts seek to build up as many followers as possible, and the number of followers does not necessarily indicate influence. Another measure is the difference between the number of followers and the number followed:
@Oprah is a good example of this
So, I've done the same thing as above, but looking at the 'Following Differential'
The picture is now very different; the largest node is now @nikicheong - but the differences are not as high as in the Followed graph above.
We can look at those with a difference of more than 100
and more than 200
So, does this mean that @nikicheong is more influential than @victorliew? Not necessarily: although it is a feature of spam accounts to have almost the same number of followers as followed, it may also just mean that @victorliew follows people back as a matter of courtesy, or that he is not very familiar with Twitter, or that his own personal networking focus is not centred on Twitter. You can't know without more investigation. It does however suggest that if he follows you, he's not necessarily interested in what you're saying
It's probably also worth noting that @nikicheong is also a columnist for a major daily newspaper, a blogger, an actor, and I dunno what else
The point being, he is likely to be active in more than one network of people, and thus gather followers from different walks of life. He is an 'Outlier' on this network, but his high 'Following Differential' suggests that people are listening to him in different networks, or clusters of social activity, too.
Conversations
Finally, another way of looking at the SMCKL online twitter network, is to look at those who actually talked to or mentioned each other. Just like on Facebook, and with blogs, many people are not very active even if they have an account.
So this is the network of people who talked to/about each other, with the node size based on Followers
And the same with the node size based on the 'Following Differential'
As you can see, it looks very different! The network is much less dense, and is more likely to reflect (to some degree) actual interactions on the evening itself. @nikicheong is in the centre here, suggesting he tweeted a lot and others responded to him. @victorliew, however, was not as active.
OK, that's it for now - any thoughts?
Now, we have all seen 'Followers' like this
Spam tweet accounts seek to build up as many followers as possible, and the number of followers does not necessarily indicate influence. Another measure is the difference between the number of followers and the number followed:
@Oprah is a good example of this
So, I've done the same thing as above, but looking at the 'Following Differential'
The picture is now very different; the largest node is now @nikicheong - but the differences are not as high as in the Followed graph above.
We can look at those with a difference of more than 100
and more than 200
So, does this mean that @nikicheong is more influential than @victorliew? Not necessarily: although it is a feature of spam accounts to have almost the same number of followers as followed, it may also just mean that @victorliew follows people back as a matter of courtesy, or that he is not very familiar with Twitter, or that his own personal networking focus is not centred on Twitter. You can't know without more investigation. It does however suggest that if he follows you, he's not necessarily interested in what you're saying
It's probably also worth noting that @nikicheong is also a columnist for a major daily newspaper, a blogger, an actor, and I dunno what else
The point being, he is likely to be active in more than one network of people, and thus gather followers from different walks of life. He is an 'Outlier' on this network, but his high 'Following Differential' suggests that people are listening to him in different networks, or clusters of social activity, too.Conversations
Finally, another way of looking at the SMCKL online twitter network, is to look at those who actually talked to or mentioned each other. Just like on Facebook, and with blogs, many people are not very active even if they have an account.
So this is the network of people who talked to/about each other, with the node size based on Followers
And the same with the node size based on the 'Following Differential'
As you can see, it looks very different! The network is much less dense, and is more likely to reflect (to some degree) actual interactions on the evening itself. @nikicheong is in the centre here, suggesting he tweeted a lot and others responded to him. @victorliew, however, was not as active.
OK, that's it for now - any thoughts?
Trackbacks
Social comments and analytics for this post
This post was mentioned on Twitter by julianhopkins: Blog post http://bit.ly/c3sxfD : social network analaysis visualisation of #smckl . Or, why @nikicheong was the networker of the night
This post was mentioned on Twitter by julianhopkins: Blog post http://bit.ly/c3sxfD : social network analaysis visualisation of #smckl . Or, why @nikicheong was the networker of the night
Weblog: uberVU - social comments
Tracked: Jan 30, 04:08
Tracked: Jan 30, 04:08
Friday, January 29. 2010 at 12:33 (Homepage) (Reply)
The Following Differential is definitely more accurate assessment of the influence a person is to a particular topic.
This graph is particularly interesting:
http://julianhopkins.net/uploads/pics_1001/jh_pic_100129_smckl_FF_200a.jpg
as you can see the people closely involved in smckl like redsheep, spinzer, zhiq and cerventus are in the center while the others fell outside.
I am not too sure why I am in the graph tho, because I only tweeted one #smckl tweet. Is the quantity of the tweets have any weightage in the node size?
Thanks.
yk
Friday, January 29. 2010 at 12:58 (Homepage) (Reply)
The quantity of tweets is not taken into account here, though it would be possible to do that too, in theory.
You're in the graph because your Following Differential is >200 - you have 266 more followers than followed.
Saturday, January 30. 2010 at 00:03 (Homepage) (Reply)
Which draws us to a potential improvement - that perhaps using followers/following ratio might be a better measure? A node with followers/following ratio of 2 says that he is a lot more influential and a far more discerning Twitter user.
A mere look at "Following Differential" is one-sided and places emphasis only on the difference between 2 absolute numbers. It places no importance on the node's taste - which you have rightly pointed out here -
"""You can't know without more investigation. It does however suggest that if he follows you, he's not necessarily interested in what you're saying """
The shrew you may already have came out with a more comprehensive solution - Looking at both the follower/following ratio (represented on the node as the size of the node) and the absolute follower numbers (nodes with followers less than X are left out) together. That may draw out a more meaningful chart.
At JamiQ, we have been studying SNA for a very long time. While we find the charts produced immensely interesting, intriguing, and impressive, we have found it hard to make immediate business decisions of it. This is the primary reason why we are still leaving SNA out of our fully automated solution. We continue to study it and innovate upon it with various partners. Once we hit upon a satisfactory solution, you will be one of the first to know.
If you want to discuss further, feel free to email me personally (kelvin.quee@jamiq.com) or the company at questions@jamiq.com
Saturday, January 30. 2010 at 16:50 (Homepage) (Reply)
Yes, you have some very good suggestions there. I had thought of the follower/following ratio, but that puts someone with 10 followers and following one, on the same level as someone with 1000 followers and following 100.
Factoring in the total followers may deal with that though I guess.
The absolute number in the Following Differential is relevant though - the Oprah example being an extreme one. Perhaps a combination of ratio and Following Differential would be better?
I agree that SNA charts need proper interpretation, but then again, so does all data. In the first instance, I see SNA as being very useful for identifying clusters for further analysis.
I would certainly be interested to know what you come up with at JamiQ, and many thanks for your detailed feedback!
Friday, February 12. 2010 at 13:45 (Homepage) (Reply)
In other words, the influence one has as a fried is different from the influence one has as a virtual friend in the Tweeter or Facebook sense.
Some say that there is no such thing as a friend on Tweeter or FaceBook. There are only electrons and a reflection of self. Tweeter or FaceBook simply provide a mirror of self having nothing to do with real community involvement.
So when one says that he or she has 2000 friends on Tweeter, one has to be careful about drawing too many conclusions relative to influence based on such a number.
What say you?
Thursday, February 18. 2010 at 13:50 (Homepage) (Reply)
One's 'Friends' on Facebook may be acquaintances, colleagues, friends, or all three. There's no reason why you can't become friends with someone online, but most of social network sites are about maintaining 'weak ties' - i.e. people who we would like to keep in our general awareness, maybe have a drink with occasionally, but we're not going to lend money to them.