Social networks – A data oriented / retro approach

This approach forms a key part of my thinking .. any comments welcome.

The Web is organised around content – for instance, most people start with ‘Search’ – when they interact with the Web. In contrast, the social web is organised around people (particiants in a social network) – and we often start by searching for specific people when we work with the Social Web.

In a nutshell, participants in a social network join the network, publish their profiles, add some content on the profiles (for example – pictures). They then create links to other users. Over time, users with similar interests will form groups where they share content. The social network so formed is a basis for maintaining social relationship, finding more content, finding other uses etc.

With the rise of Facebook, MySpace and other sites, Online social networks have become popular. However, social networks themselves have existed for a long time now (both online and offline) – even before the Web.

In fact, any body of transactional data derived from participants in an electronic conversation can be viewed as a social network (such as email data, telecoms call records, Instant messaging data, forum posts etc)

When viewed in this way, the two common elements which underpin social networks are; The conversation between users and The underlying data. This is especially relevant in a converged scenario (Web, Telecoms/mobile and Media) – where conversations may span platforms.

In fact, a social network can be viewed as an ‘umbrella’ layer on top of any conversational data.

This approach is similar to Overlay networks . An overlay network is a network built on top of another network. Nodes in an overlay network can be seen to have a set of logical (virtual) links between the nodes of the base network. For instance, a Peer to Peer network can be seen as an overlay network created on top of the Internet. Overlay networks reveal new pathways and relationships between the nodes of the base network and are a reflection of the usage of the nodes. Thus, depending on function, multiple overlay networks can be built on top of the base networks – a social network for business, a social network for dating etc etc.

Whichever way we look at this, underlying all the networks lies raw data.

The data and insights derived from popular social networks like Facebook and MySpace has given us many new insights into user behaviour – all of which can be quantified.

This means, we can take a ‘retro’ approach i.e. i.e. we can apply these new insights from social networks to an existing body of conversational data. This approach is universal because every organization has access to such data from the many touch points via which it engages with the customer.

However, social relationships are not always explicitly detectable from data; hence it is more important to derive relationships between nodes from the underlying data rather than depend explicitly on the information provided by the nodes themselves. For instance, profiles are often incomplete and are based on easily entered or default information. In contrast, if a relationship can be derived between two nodes based on transactional information, it is more insightful.

This approach applies to many scenarios(from Telecoms Operators, to social networks to airlines) and it has many uses ranging from managing trust, combating SPAM, viral propagation of content, managing reputation, social media marketing etc.

Despite it’s universal appeal it is based around only two core tenets: Underlying data and the conversations. It is platform agnostic and can apply to any company which has a body of data from customer touchpoints(Telecoms Operators, media companies etc). It is relevant since we can retrospectively apply many new insights from social networks like Facebook to existing body of data

If you are interested in discussing this(even the mathematics behind the approach), then please email me at ajit.jaokar at