Archive for the ‘networks’ Category
Newer Entries »Kerry’s map of the top 50 twittering journalists
Wednesday, January 7th, 2009
My colleague, Kerry Gaffney, has just posted her analysis of the network formed by the top 50 UK journalists on Twitter.
She says:
Looking at the original map, it immediately seems obvious that the PR bunnies of the world are far more likely to link to each other, but just to make sure we dropped both datasets through UCInet and looked at the density scores, and sure enough the PR network is almost twice as dense, sharing 1459 ties compared to 785 for journalists. Or a ratio of .595 against .320 for following within the group, so not quite double, but not very far off.
If you’re interested in this sort of thing (and who, these days, is not?) then I recommend that you take a look at Kerry’s analysis.
Tags: journalists, kerry gaffney, mapping, maps
Posted in networks, twitter | 1 Comment »
Some Twitter Social Network Analysis
Wednesday, December 17th, 2008
On November 10th, Stephen Davies collected together a list of “UK PR people on Twitter” According to PostRank, this (and his earlier post, “UK Journalists on Twitter“) are the most popular posts on his blog.
Then a couple of days later, Stephen Waddington pushed that list through TwitterGrader to come up with his list of “Top 50 UK PR people by Twitter influence“
A couple of weeks ago, I was looking for a seed list with which I could test our “whitelist” and “canonify exception” rules on Rufus (the network analysis tool that Porter Novelli has been working on for the past six months.) This isn’t the right place to go into it, but to put it simply, the whitelist restricts the search to domains that are on the list (like a guest list), and the canonify exception list stops Rufus from chopping the subdomains or directories off the list (without this, a site like sethgodin.typepad.com would just show up as typepad.com or en.wikipedia.org/wiki/Social_network_analysis would show up as wikipedia.org. Rufus, by the way, is named after the George Carlin character in Bill & Ted’s Excellent Adventure.
My colleague, Tim Hoang used to work with Stephen W., so he sent him the image. Wadds then posted “the map on his blog“. My flickr page has never had so much activity.
Here’s the original graph:
Lots of people started drawing conclusions about the nature of PR, or the nature of Twitter from the graphs. There was lots of interesting speculation. Some people thought that this demonstrated how introverted the twitter crowd is. Others thought that it showed how introverted the PR/Social media crowd is. Others seemed to think that it didn’t matter.
(more…)
Tags: citation analysis, graph, influence, influence mapping, network analysis, networks, research, social media, twitter
Posted in influence, networks, twitter | 29 Comments »
Relationships between “top 50″ UK PR twitterers
Tuesday, December 16th, 2008
This is a 300dpi map of the top 50 PR twitterers (as per Stephen Waddington’s analysis) and the interrelationships between them.
To generate this:
We first crawled all the accounts for “friends” (accounts that they follow) and “followers” (accounts that follow them). This is a profligate use of resources because we were always going to throw away a massive load of that data. But it’s always more interesting to start with a large data set. You don’t know what you’re going to find.
Then I wrote a quick-and-dirty perl script to process the data looking only for those instances where one of the top 50 followed another.
Then we dropped everything into NetDraw (if you are at all interested in this stuff, you really should get hold of a copy and start reading around the subject.) We laid out the chart so that the people who have the most peer-group followers are in the centre of the chart – and to make it even more obvious, we sized their nodes according to the number of peer-group followers that they have.
So people on the peripheries (like me – mediaczar) are peripheral to the community, and those in the middle are central. Obvious, huh?
This chart already shows a massive difference between our analysis (as it progresses) and the raw data from Wadds’s list. There are some really good reasons for this, which I’ll go into on the blog.
Tags: citation analysis, influence, influence mapping, network analysis, networks, twitter
Posted in influence, networks, twitter | 4 Comments »
Map of top 50 UK PR twitter people and their followers
Saturday, December 13th, 2008
This is not a hedgehog in a cranberry field. It is a network map, but a particularly tightly-knit one.
Spurred on by some of the comments we’ve received about the Rufus map we made of the top 50 UK PR twitter people (as measured by Stephen Waddington) I thought it’d be a good idea to look at this in a bit more depth. Rufus isn’t really the right tool for looking at this kind of thing, so we’ve built something else to do it better. Looking at one site or service is a lot easier than looking at lots of sites — so this took hours, not months to create.
After a little debugging we were ready to test on a seedlist of 50. The crawl took about an hour to run.
This is a visualization of the data set we got (correct as at December 12, 2008) after very little processing.
The size of the blobs relates roughly to “how many people in the group follow you.”
We’ve removed anyone who is only followed by one person in the group. So everyone here is followed by at least two others (obviously.)
There are just too many people in this graph to show labels. And a lot of the top 50 people are hidden by other top 50 people. Maybe I should do a graph rotating in 3D. (Later, having tried this: if I had a SGI workstation, maybe I would.)
What I’ll do over the weekend is process the data files I’ve got (one’s got around 30K records, and the other 40K records) to see if we can tease a little more information out of them.
Then we’ll run this on other twitter communities, and random twitter seedlists to see how (if at all) the networks differ. Are PR people more introspective than the rest of the twitterverse?
This is a very high def image, so it will blow up nicely.
Tags: citation analysis, influence mapping, mapping, network analysis, networks, public relations, twitter
Posted in influence, networks, twitter | 2 Comments »
High network density in twitter UK PR community
Monday, December 8th, 2008
For this graph we took a list of the top 50 PR twitterers as measured by Stephen Waddington (Nov 2008). We added “twitter*” to the canonify exception list to identify individual twitterers (this isn’t perfect — the regex may need some tweaking) and further limited the crawl to domains that contained the word “twitter” using the Whitelist function.
Again – look at how dense the network is here.
Tags: network analysis, network graph, rufus, twitter
Posted in influence, networks, porter novelli, rufus | 1 Comment »
Reading RUFUS data with yEd
Monday, December 8th, 2008
Most of the time we use UCINET and NetDraw to analyze the data from Rufus. Rufus exports crawl data to a Pajek .net file by default. But we can also export GraphML and read the data into other tools that handle that format. This is a test we ran of this feature using yEd
It’s not working beautifully yet, but it is working.
Tags: graph, network analysis, rufus, yEd
Posted in influence, networks, porter novelli, rufus | 2 Comments »
A sneak peek at our online influence mapping tool
Thursday, December 4th, 2008
Porter Novelli has been working on its own “online influencer mapping” tool for about six months now. Recently, I’ve started posting screen grabs on our Flickr page to see what people think about it. I thought it was probably time to share some of the images here.
Version 3.5.4 (Always in Beta)
The project is named Rufus after the character George Carlin played in “Bill & Ted’s Excellent Adventure”.
For those of you who know how network analysis works and what it’s used for, this is revolutionary only in that it’s fast and accurate enough to use as an exploratory tool.
For those of you who have no idea what network analysis is or how it’s used in many, many situations, 2009 would be a really good year to start finding out.
For this graph (which took around 5 mins to generate), we took as a seed list the first 50 back links as generated by Yahoo Site Explorer (http://siteexplorer.search.yahoo.com/.) We’ve tested this up to 100 seeds, but there’s plenty of room to go further.
Tags: citation analysis, influence, influence mapping, network analysis, networks, porter novelli
Posted in influence, networks, porter novelli, rufus | Comments Off
Mapping the social graph of weight loss groups
Wednesday, July 2nd, 2008
These are the graphs from some research on weightloss groups on Facebook. I’ve processed the data so that:
1) the size of dot is related to "total number of friends" – this only works where a user’s friends are publicly visible – quite often they aren’t, and I haven’t checked to see what the incidence of this privacy setting is generally and specifically
2) all isolates (i.e. those users who have no (public) personal relationships within the group have been removed.
This is the network graph of relationships on a personal weight loss support group. A college student set this up to support her own goals. She told me: " For my group, I just started it out by inviting all of my friends and then some people joined the group who found it in a search, I think. I am amazed by the amount of support I receive from random people who encourage me to keep on going. There are some spammers on the group who are just there trying to sell stuff and that gets annoying, but I know I can’t avoid them."
This is the network graph of relationships on an unofficial weightwatchers group on Facebook. You can see that there are hardly any member-get-member relationships here. My friend Valery (who has a professorship in this sort of thing at Wharton) says:
"It’s very common that organizations and interest groups become foci for personal networks. In fact, I believe that joint activities are the prevalent mechanism of tie formation. "
But it doesn’t look like it here. Looks to me that – while people may form relationships around special interests – they don’t mirror these on Facebook. Say I suffer from Meniere’s Disease (apparently true) and I participate in a Meniere’s support forum (not true at present), I don’t necessarily make those people my Facebook friends…
Another example of the "not many personal relationships" graph for a weightloss support group on Facebook.
How do people get information on weight loss? After a few interviews, I think the answer is like this:
1) Influencers are "pull", rather than "push" resources (I’m thinking of going on a particular product, so I mention it casually to several friends to gauge consensus/temperature. One or more of them tell me "oh yes, I’ve heard of that", and one tells me "yes, My friend tried that, and lost 20lbs") This is not an active market. Most people won’t be evangelizing, and evangelizing behaviour may even appear suspicious.
2) That said, people trust strangers to an extraordinary degree. Friend-of-friend endorsement is readily accepted, as is the anonymous commentary on boards & groups. Bloggers are slightly less trustworthy, it seems – because most of them have an axe to grind.
OK — so this really isn’t v. scientific. But compare this to the map of green issue member-get-member activity and you’ll see a huge difference.
Tags: facebook, influence, mapping, networks, research, social media, weight loss
Posted in networks | 4 Comments »
Unravelling member-get-member activity
Tuesday, June 17th, 2008
This is a network map. You’ve probably seen something like this before. What’s quite interesting is that this is an undirected graph (that is, the links go both ways – if A knows B, B knows A, that sort of thing) but there’s some directed activity implied.
Groups on sites like Facebook and Bebo work by members recruiting other members, either actively (“join this group”) or passively (friends can see on their friend feed that other friends have joined up, and decide to get involved).
We’d expect to find that people have strong existing personal relationships within such groups.
We took one group (a green interest group) which has around 2,000 members. We selected this group because it is has a clear single-minded proposition,as well as a strong local element that means that people are more likely to know each other than on some of the more generalist boards.
There are more than 250K publicly available relationships inside and outside the group, and we looked at analyzing all of them.
It seems that there are a few highly connected people (people with between 395-400 friends) who sit at the centre of these things, and on whom the eventual success of the group depends. While they may be no more active than other users (and may even be less active), they are the hubs who link together the network.
In the maps you can clearly see three kinds of shape.
Long fingers show where users get users (usually one or two at a time). Fan shapes show one user who mobilizes many friends. These are clearly interesting to us. In the middle are “mares nests” where lots of people know each other.
This is a pretty straightforward Pareto-like distribution: 33% of the users are connected to 79% of the members, the remaining 66% only link to 22% of the users.
Here’s how we’re going about it.
1) Spider every member of a group, and their friends
This shows the members of a group (red) and all their publicly-visible friends. This is a much smaller group than the one in the graph above (these things grow exponentially, you understand)
Red dots are members, grey dots are their friends.
I’ve gone through and blurred out the individuals’ names. We probably shouldn’t have been collecting those anyway…
2) Look for relationships between members
Here we just use a little perl script to sort through the lines selecting only those relationships between two members. In essence we’re throwing away all the other relationships. Perl is good at this sort of thing. I found a trick over at the perlmonks site that shows you how to search an array much faster than just grepping (which took a while).
3) Drop the results into NetDraw
We’re using Analytech’s excellent/free network drawing application, NetDraw. You can tweak the Windows Metafile output in things like Illustrator.
Tags: mapping, networks
Posted in influence, networks | 4 Comments »
