Mining Social Software Interactions to Develop the Enterprise Social Graph

In doing some research, we came across an excellent white paper by Laurence Lock Lee, James Matheson and Stewart Mader. In WikiMining, WikiNetworks and Wikinomics, the authors studied the co-editing and commenting interactions of participants in the wikipatterns.com wiki community.

Objectives of this research included:

• Measuring the health of the wiki community,
• Understanding which participants had the highest network centrality, and
• Seeing how well this interaction analysis correlated to the actual social relationships among participants

We've written previously about social network analysis and its value in a corporate setting. While the wikipatterns.com research is not in the enterprise, much of its analysis is consistent with Professor Rob Cross's work on employee networks inside organizations. Read the paper for some good real-world context to the principles of social network analysis.

What especially caught our eye were these two findings:

1. Correlation of inferred social networks to surveyed actual social networks
2. Percentage of relationships that occurred strictly because of the wiki

Let's examine these two findings a bit more.

Correlation to Actual Social Networks

The study's authors mined the editing and commenting interactions among the wiki participants. The idea here is that if you and a colleague both edit a wiki page, that is tallied as a "relationship unit" between the two of you. By analyzing the depth and breadth of these relationships, the authors constructed a social graph among the participants. These map of the social graph, called a sociogram, is shown below:

 

The authors then surveyed wikipattern.com participants, asking them to identify their actual social relationships within the wiki community. Only a subset of people responded, but enough did to construct a meaningful stated social network for participants.

The authors then compared the inferred social network to the stated social network. The basis of comparison was "degree centrality", which analyzes the number of connections a person has. The table below shows the Top 5 participants based on degree centrality:

 

In looking at the top five most central participants, four names show up in both lists. In reviewing all participants, the authors find a statistically significant 0.77 correlation.

What this says is that there is likely a way to use social software, such as wikis, to mine the enterprise social graph. The study authors are appropriately cautious with this declaration. Their study is one of the first we've seen that tackles this idea.

Social Software Fosters New Relationships

An important benefit of social software is the diversification of one's relationships, which we've discussed here before. Finding and interacting with others outside of your regular internal contacts has tremendous value for you personally, and for your company overall.

In conducting the wikipattern.com user survey, participants were asked to characterize the relationships they had with others on the wiki. There were four options:

• Prior relationship > 2 years
• Prior relationship < 2 years
• Relationship created through the wiki but now extending beyond it
• Relationship through the wiki only

The survey participants characterized a totla of 117 relationships based on one of these four options. Of those 117 relationships, 37 were the latter two types - formed through the wiki. In other words nearly a third of the relationships were based on participating in the wiki.

This finding is valuable. It shows that providing an open collaboraiton and sharing platform results in new connections not previously made. Which is a key benefit of Enterprise 2.0.

Looking forward to more research in this area.

Comments

[billycripe]

It's a great piece. Leveraging implicit or organic social networks that emerge from other activities will is the next step. I published about doing this with email last September in Computer Technology Review http://tinyurl.com/csage2 and in chapter 10 of my book, "Reshaping Your Business With Web 2.0" (McGraw-Hill 2008). It is really nice to see study / statistical validation of the concept.

[Michael]

The blog is getting spammed (3 of 4 comments here are spam) and Connectbeam does not answer mails or phone. Do anybody know what is going on?

[Force Factor Supplements]

Really a educative and informative post, the post is good in all regards,I am glad to read this post

[social bookmarking]

good work
bravo