Communities and Connections: Social Interest Mapping
Beyond the proof that Stephen Fry really does exist at the heart...
Beyond the proof that Stephen Fry really does exist at the heart of Twitter, what can we learn from exploring interrelations between groups of Twitter followers?
- Duration: 10 mins
- Published on: Thursday 20th September 2012
- Introductory Level
- Posted under: Technology
The themes of this week's TEDxMK event of Communities, Connections & Conversations invite participants to explore how technological and social innovation can strengthen and enrich public life.
The rapid growth of online social networks will surely play a part in this, providing as they do a way for individuals to congregate and communicate based on personal, professional, commercial and interest led relationships.
To a certain extent, today's dominant social networks may be caricatured (at least on the basis of their origins), in the following ways:
Facebook has captured what is often referred to as "the social graph", a network of connections between people who are related, know each other personally, or who come to know each other as friends of friends.
LinkedIn is a social network based around company connections: people who link to each other tend to work for the same company, have done business with each other, or work in the same industry. We might say that LinkedIn is building of view of "the professional graph".
Twitter, as it grows rapidly in public consciousness as the place (or "backchannel") to chat around live events as diverse as television and radio broadcasts, conferences, and "trending" hashtags, increasingly provides a snapshot of "the interest graph".
One of the interesting things about Twitter is that, for the moment at least, it makes friend and follower connections between all non-private accounts public.
That is, given someone's Twitter username, and whether or not I personally follow them or they personally follow me, I can grab a list of who all their friends and followers are.
I don't (currently) even need my own Twitter account to retrieve this information. (Things are changing though, as Twitter starts to take increasing control over its data...)
If we take follower links on Twitter as signs that someone is somehow interested in the accounts they follow, and if we also have some idea of what particular Twitter accounts might be notable for (for example, @OpenUniversity communicates information about the OU, @ComActMK represents the community action and charitable third sector in general, and in Milton Keynes in particular, @stephenfry is about whatever Stephen Fry finds quite interesting...), we can start to create social interest maps based on accounts that are commonly followed by people who are interested in particular topics.
Here are a couple of ways we can generate these maps:
The followers of@ComActMK are presumably interested in the sorts of thing that @ComActMK tweets about. Those followers presumably also follow other accounts that are in the same interest area, at least in part. (I suspect quite a few @ComActMK followers might have other interests too, "outside" the community action area, such as Formula One motorsport, or watching XFactor.)
The image below shows a map of accounts that are commonly followed by the followers of @ComActMK. The labels are sized (broadly speaking) according to how well followed each account is by the followers of @ComActMK.
The names are positioned using a particular layout algorithm that tries to position names that are similarly connected close to each other. In this way, we can see different "special interest" areas evident within the map if we associate particular Twitter users with particular interests.
In this map, we see three broad communites - one to the left, one in the middle and one to the right. The areas are coloured using a "false" colour applied to three groups of users detected using an algorithm that tries to identify sets of users that are heavily connected to each other within a group and less connected to users in other groups. So for example, if you have a bunch of people at an annual "local sports" festival, the football players might know a lot of football players, the cricketers might know a lot of cricket players, and a few cricketers might know a few footballers, or actually be footballers, but we'd broadly speaking colour people as either footballers or cricketers.
In the @ComActMK map, the group to left appears to be made up of groups that are active in Milton Keynes. If you zoom into the image, you'll see a lot of the usernames on the left side of the map include MK in the username.
If you now drag the image so that you can inspect the right hand side of the graph, (the purpley blue area), you'll see a lot of user names that appear to relate to charitable causes.
The group in the middle appears to be be rather more eclectic. To help us make sense of the maps a little more, the word clouds are generated from the Twitter descriptions associated with each Twitter user in a particular group.
The large words appear more frequently in the descriptions of users in each group, so we see that the group to the left does indeed appear to have a strong Milton Keynes representation, the group to the left does indeed appear to signify the charitable and voluntary sector interest area, and the group in the middle is a bit of a general mix - the sort of things "everyone" is interested in.
One way of doing this a little more rigorously is to look at the personal descriptions of each of the Twitter users in a special interest area and then generate a word cloud from those descriptions.
And here's something else that is quite interesting.... If you peer closely at a large number of social interest maps, you'll find that whatever the interest area, @stephenfry appears.
This presents us with something of a challenge: trying to identify accounts that are disproportionately followed by one audience or set of users compared to another; that is, to identify accounts that sit above some sort of "background interest" level. This is, needless to say, an active area of research.
In a sense, mapping out the people commonly followed by the followers of @ComActMK allows us to identify the emergent social positioning of the Twitter user in an interest spaced defined by the expressions of interest the followers of @ComActMK reveal through their public following of other Twitter accounts.
This act of collective intelligence provides a way of mapping out interest areas that has previously been hard to describe at such large scale.
As well as generating emergent social positioning maps around a specific account, we can also generate social interest maps around particular areas of discussion, such as a hashtag, or a shared link. By searching Twitter for mentions of a particular tag or link, we can get a list back of people who have recently used that tag, or shared that link.
Using this set of people in much the same was as we used to set of followers of a particular Twitter account (@ComActMK in the above case), we can generate a map of the common interests of the users of the tag or the sharers of the link.
So for example, if we search for people recently mentioning allotment and gardening, and then for each one grab the list of people they follow, we can see who people who recently chatted about allotment gardening tend to follow - that is, what sorts of interests they might have:
Generating these maps is one thing, of course. Learning how to use them to foster community action and conversation is another. If you have any ideas about how you might be able to use such maps in your own interest area, or would like to see an interest map generated around a particular Twitter account or recently used hashtag or search term, please leave comment below, and we'll pick out at least one lucky winner for social interest mapping treatment :-)
P.S: If there is enough activity around the #TEDxMK hashtag over the weekend, we'll post a map of the social interest map around it...
Copyright & revisions
If you enjoyed this, why not follow a feed to find out when we have new things like it? Choose an RSS feed from the list below. (Don't know what to do with RSS feeds?)
Remember, you can also make your own, personal feed by combining tags from around OpenLearn.