How People Gather

If I give you a phone under the condition that I get to track every person you interact with using GPS, WiFi, Bluetooth and much more and you have to fill out a complex survey revealing very personal details about your self. Would you do it? Well that is what about a thousend students have done, for a study in Copenhagen aming to find a better way to analyze social networks in.

The strategy was to look at how the connections between people varied every 5 minutes and over time use something called the Jensen-Shannon similarity to identify which of the groups were the same. The only thing you need to configure is the similarity threaschold. Then you could suddenly identify people with a partner, family members, social friends, co-workers. Just by looking at when and where they were meeting. One of the more entertaining examples was a party that ended with a few couples going home at three ó clock and spending the rest of the weekend together.


Here you can see how people move about during the day.

Some of the tools used to do this include the map equation that you can read more about at http://www.mapequation.org/apps/mapdemo.html and for the graphical something that looked like this https://marmelab.com/EventDrops/ and this http://orgo.stolarsky.com/ was used to represent the data visually. They are both found in the collection of graphical tools at https://d3js.org/.

The soure code can be found at https://github.com/mapequation/infomap if you want to try the entire application on your own data.

- Socal networks, infomap, Jensen-Shannon similarity

<< The AI revolution in Science Action in mind >>

Comment

Name
Mail (Not public)
Send mail uppdates on new comments

Comments

1 post found

Hey! Thanks for writing this. Here's a link to the visualization that I demoed during my talk: http://ulfaslak.​com/​research/​temporal_communities/.

2017-11-23 11:26:05 - Ulf Aslak