Home > Datasets, what i've been doing > Social Sensing Dataset

Social Sensing Dataset

I finally got around to extracting out contact events from the social sensing study. I have made some notes on the Social Sensing Dataset page. The main images are shown below:

Social Sensing Dataset activity, 15mins, hourly, daily, 3 daily, and weekly. Lines showing the most active period

Social Sensing Dataset activity, 15mins, hourly, daily, 3 daily, and weekly. Lines showing the most active period

It seems most active from Oct 2008 onwards, and tails off, with the end of Nov 2008 being a good cut-off point. This means there is over a year of contact data.

Force directed connected time graph for SocialSensing dataset, showing mapping of users to nodes, with low weight edges removed. NOTE: Do not share this image - remove names/btmacs first

Force directed connected time graph for SocialSensing dataset, showing mapping of users to nodes, with low weight edges removed. NOTE: Do not share this image - remove names/btmacs first

The connected time graph (i.e. edges weighted by the total length of time connected with the target node) gives a really good sense of the social structure. As I personally know most of these people, (and I am on it myself) it is easy for me to see the relevance of the layout.

I haven’t written code to lookup locations based on cell towers/wifi points yet, but it might be interesting in the future if we end up using location.

Also, I need to work on getting BTMAC addresses for the three other people. The problem is that I know that at least one of these people changed phones during the study – hence why they are removed.

  1. Pádraig
    June 16th, 2011 at 08:06 | #1

    This is a very interesting diagram. It should be good for BubbleH as well as there is good hierarchical structure there. It will be interesting to see what the eval shows.

  1. No trackbacks yet.