The last few years have seen the adoption of social networking increase rapidly. From Facebook to Twitter, LinkedIn to Flickr – there is a social network for just about anything.
As the revolution of social networking continues unabated, there comes a growing need to explore patterns within the networks – a process called social network analysis (SNA)
Previously, the world of social network analysis could only be accessed with a bit of computing knowledge. However, an open source programme called Nodexl has changed that by bringing some of the important metrics used to understand a network, and the ability to create impressive network graphs, into Excel.
Nodexl makes understanding a social network graph easy for anyone who can navigate around a spreadsheet. Excel is often where the world of computer programmers and the rest of us can meet up and speak the same language. Nodexl also makes it easy to import data from existing social networks such as Twitter, Flickr and Youtube
The people that can begin to make use of network graphs range from marketers to activists – and I imagine they are now a staple of any well equipped social media political campaign. Using a social network graph you can (among other things):
- Spot the trusted influencers in a network
- Find the important people that act as bridges between groups
- Uncover isolated people and groups
- Find the people who seem good at connecting a group
- Plot who is at the centre and who is at the periphery of a network
- Work out the where the weakest points of a network are
- Assess who is best placed to replace a network admin
- Node: In a social network a node will usually represent a single person – but it can also represent an event, hashtag etc
- Edge: A connection/interaction between two nodes – such as a friendship in Facebook, a follow on Twitter or an attendance at an event or Twitter Hashtag.
One major question that a social network analysis asks is how connected nodes (or people) are. But what determines how connected any person is? What metrics can be used to work it out how influential or powerful any individual player is?
These are some of the major metrics used in Nodexl – and they offer a good way to start thinking about your own networks:
- Centrality – A key term which refers to how ‘in the middle’ a node is in a network.
- Degree centrality – a count of the number of nodes a node is connected to. This could be the number of people that follow you on Twitter, or the amount of people that viewed a YouTube video. It is important to remember that a high degree score isn’t necessarily the most important factor in measuring a nodes importance.
- In Degree and Out Degree – A connection between two nodes can be undirected (we are mutual friends on Facebook) or directed (you follow someone on Twitter that doesn’t follow you back). The In-Degree refers to the number of inbound connections, and Out-Degree refers to the number of outbound connections.
- Geodesic distances – A geodesic distance is the shortest possible distance between two nodes (popularly known as the degree of separation). In social network analysis, a nodes shortest and longest geodesic distance is recorded (the longest possible distance between a node and another is sometimes refered to as its eccentricity and can be used to work out the diameter of a network). An average geodesic distance of an entire network is worked out to assess how close community members are to each other.
- Closeness centrality – This metric determines how well connected a node is in the overall network. It takes into account a nodes geodesic distance from all other nodes. Using this metric you can find people that don’t have strong connections.
- Betweenness centrality – A score of how often a node is on the shortest path between two other nodes. This can be thought of as a bridge score – how important a node is at bridging other connections. People with a high betweenness centrality are often known as key players. A node could only have a degree centrality of 2, but if those two connections bridge to large unconnected groups, then that node will have a high betweenness centrality.
- Eigenvector centrality – This looks at how well connected the people you are connected to are. It scores how much of a network a node can reach in comparison to the same amount of effort enacted by every other node in the network.
I am going to be exploring social network analysis over the next few weeks and blogging what I find here – if you want to follow along make sure you follow me on twitter or subscribe for updates.