Science and Industry

As discussed previously, Juuso Karikoski and I made a tiny study of multiple social networks. And we got trough the peer-review, and thus headed to IEEE Social Computing 2010.

In this post¹, I focus in the two interesting keynotes, first by Dr. Nashir Contractor and the second one by the famous Dr. Alex Sandy Pentland.

World of Warcraft, Katrina and multilevel networks

Nashir Contractor spoke how networks could be seen as multilevel — instead of just people (or objects) networking and generic links, there could be social objects and people mashed together with meaningful links: I could be the author of this blog text.

Nashir also spoke about computational social science² — a field where both computer sciences scholars and social sciences scholars need to collaborate.

However, he had some more actual content too. They’ve isolated eight different kind of reasons why we network:

• self-interest
• social and resource exchange
• collective action
• social status
• balance
• homophily
• proximity
• co-evolution

And guess what. Based on their research we can distinct these patterns based on structural behavior. They’ve actually used this method in Science of Networks in Communities-research group to detect gold farmers in EverQuest 2.  They’ve also examined the hurricane Katrina and pinpointed the time when FEMA’s role in relief efforts failed. Thus, the question rise: what if we had this kind of tools every day to help us make better decisions.

How social networks make us smart?

The topic of making these part of our every day toolset was continued by Sandy. His approach was a bit more technical: putting sensors every where. And what can we get out of that?

The first examples came from organization studies: the project where one team was only contacted via email and others regularly meet face to face failed nicely. And, these kind of tools allowed visualization of the information flow inside the company. These technologies are now tested also in the Army to make their work more efficient.

However, from that we moved to more urban sensing, maybe something I would call city movements. With this amount of technology we can pinpoint where you live and move. First nice result from that is smart traffic applications, busses and taxis being in right locations. However this gets more deeper, as Sandy explained.

Based on where you live and move, they actually could estimate weather you had an iPhone or an Android, and all sorts of strange data. We went trough bunch of colorful maps and it almost got a bit scary. And, as said: this data is already now easy to collect.

But as said, this was scary, and privacy issues were discussed. Let’s see what actually happens in data ownership in few years…

1. and some comments on papers will follow later… [↩]
2. This is defined Science-magazine and seems to be big thing in the States, severals degree programs in this area… Some of them actually sound interesting. [↩]

One thing I’ve been interested for some time is friendship in online environments. Services like Facebook or Twitter allow us to add friends or to follow, but what these actually mean is totally different discussion. I know, I’m not the only one doing this kind of things — actually, I’m behind the masters¹ in this area.

However every now and then you get interesting ideas that what could be done when data just emerges and you take a deeper look to it. The following is something Juuso Karikoski (Aalto University) and me have worked on and shall present in the IEEE Social Computing 2010 (stay tuned for a conference report), but I’ll make a human readable version out of here.

It’s even now possible to use one data source and try to understand structure in a service. Sometimes it’s unidirectional and sometimes the system allows you to weight this data. However, the problem is that it just shows one channel of communication, but we use several different kinds of channels: I communicate via IRC, Facebook, email etc. — and looking just one service gives us a limited view.

What we worked out was a solution to mash together two different kind of data sources: phone call log data and a data from an Internet service called OtaSizzle. Naturally, this view is still limited and our sample size was hilariously tiny (n << 100), but it’s more than nothing. What we basically show is difference in these two networks.

As you might guess, we are not the only ones who have this kind of results: Karien Van Cleenput has just shown similar thing using a survey, and with decent sample size too. She suggest that strong ties are maintained with full variety of communication where as weak ties are maintained using social networking sites and face to face as their means.

What these results mean for me at least then? By looking several data sources together, we can build superior experiences for the users, something that makes them smile every day. And we need to understand the complexity of human nature, how they communicate using variety of media…

1. For example, see Huberman et al (2009), Golder et al (2007) and Eagle et al (2009) for using online or mobile phones as data sources and making observations of the data and Donath et al (2004), Donath (2007) and Fono et al (2006) for explaining human behavior in online networks. [↩]

I was one of the student volunteers in Pervasive 2010 conference held in Helsinki, Finland. As a SV, I’m required to participate in the conference organizing but had the change to attend certain sessions also, luckily. It was interesting to see, what others are doing in the are of ubiquitous computing (or pervasive, what ever you want). Some highlights as part one of this…

There was an interesting video on human-pet interaction enhanced with a sensor pack for the cat. Also, as I’m a cat lover, this naturally got my eye. However, there were several similar things ongoing, namely sensor stuff and wearable computing. Maybe some day we learn to play music with the help of wearable stuff or I wouldn’t get so lost when my belt would vibrate when I’m not moving to the right direction — or then not. One of the big things I’ve learned is that things living  the lab may still be too freaky to be used in every day environment…

Secondly, as my background is in political science, it was nice to see the term citizen science in one of the papers. The idea was rather trivial: having a sensor network (yeah, old stuff) to have some data on air pollution. However, there was a tiny twist: the prototype also included a web based service, where the results gathered were discussed — and this is where the deliberative democracy just might kick in… Need to re-read that paper, definitely and check how to cite that stuff in my Master’s thesis.

And the rest will come later: it’s sunny outside and I’m sitting in the office; clearly not using all the possibilities of today’s pervasive stuff. So, off I go.