syslog
15Jul/110

Socio-spatial properties of online social networks

Posted by Salvatore Scellato

Some social scientists have suggested that the advent of fast long-distance travel and cheap online communication tools might have caused the "death of distance": as described by Frances Cairncross, the world appears shrinking as individuals connect and interact with each other regardless of the geographic distances which separates them. Unfortunately, the lack of reliable geographic data about large-scale social networks has hampered research on this specific problem.

However, the recent growing popularity of location-based services such as Foursquare and Gowalla has unlocked large-scale access to where people live and who their friends are, making possible to understand how distance and friendship ties relate to each other.

In a recent paper which will appear at the upcoming ICWSM 2011 conference we study the socio-spatial properties arising between users of three large-scale online location-based social networks. We discuss how distance still matters: individuals tend to create social ties with people living nearby much more likely than with persons further away, even though strong heterogeneities still appear across different users.

15Jul/110

Efficient Social Sensing based on Smart Phones

Posted by Kiran Rachuri

Mobile smart phones represent a perfect platform for building systems to capture the behaviour of users in the work-places, as they are ubiquitous, unobtrusive, and sensor-rich devices. However, there are many challenges in building such systems: mobile phones are battery powered and the energy consumption of sensor sampling, data transmission, and resource intensive local computation is high, the mobile phone sensors are inaccurate and not specifically designed for the purpose of capturing user behaviour, and finally, the local and cloud resources should be used efficiently by considering the changing mobile phone resources.

We address the above technical challenges for supporting social sensing applications in a paper to be presented at the upcoming ACM MobiCom '11 conference.

In the paper we describe the design, implementation, and evaluation of SociableSense, an efficient and adaptive platform based on off-the-shelf mobile phones that supports social applications aiming to provide real-time feedback to users or collect data about their behaviour.

The key components of the system are:

- A sensor sampling component adaptively controls the sampling rate of accelerometer, Bluetooth, and microphone sensors while balancing energy-accuracy-latency trade-offs based on reinforcement learning mechanisms. The learning mechanism adjusts the sampling rate of the sensors based on the context of the user in terms of events observed (interesting or not), i.e., the sensors are sampled at a high rate when there are interesting events observed and at a low rate when there are no events of interest.

- A computation distribution component based on multi-criteria decision theory dynamically decides where to perform computation of tasks by considering the importance given to each of the dimensions: energy consumption, latency, and data sent over the network.  For each classification task that needs to be processed, this scheme evaluates a utility function to decide on how to effectively distribute the subtasks of the classification between the local and the cloud resources.

We show through several micro-benchmark tests that the adaptive sampling scheme adjusts the sampling rate of sensors dynamically based on the user's context and balances energy-accuracy-latency trade-offs. We also evaluate the computation distribution scheme in terms of selecting the best configuration given the importance assigned to each performance dimension, and show that the computation distribution scheme efficiently utilises the local and the cloud resources and balances energy-latency-traffic trade-offs by considering the requirements of the experiment designers.

To further demonstrate the effectiveness of the SociableSense platform, we also conduct a social experiment using an application that determines the sociability of users based on colocation and interaction patterns. The use of computation distribution scheme leads to approximately 28% more battery life, 6% less latency per task, and 3% less data transmitted over the network per task compared to the model where all the classification tasks are computed remotely.

Kiran K. Rachuri, Cecilia Mascolo, Mirco Musolesi, Peter J. Rentfrow.  SociableSense: Exploring the Trade-offs of Adaptive Sampling and Computation Offloading for Social Sensing. In Proceedings of the 17th ACM International Conference on Mobile Computing and Networking (MobiCom '11), Las Vegas, USA. [PDF]

12Jul/110

Web Science Summer School

Posted by Daniele Quercia

till tomorrow, i'm at the web science summer school. i was invited to give a talk on privacy in mobile-social networking applications. my talk was a re-mix of blog posts and papers (including spotme, "what we geeks don’t get about social media privacy", and "location-related privacy in geo-social networks" - pdf ). unfortunately i could not attend the whole summer school, but you can check here the schedule and my notes on a couple of talks are next.

marcel karnstedt gave a great presentation on the effects of user features on churn in social networks. he presented a nice empirical study of the mechanisms by which a web forum maintains a viable user base. he found that different forums show different behavioural patterns and also found few interesting regularities. have a go at his paper (pdf)

bernie hogan wondered what kind of mental models people have of their Facebook personal (ego) networks. to answer this question, he collected mental models that a number of Facebook users have about their personal networks, collected the actual personal networks from Facebook, clustered them using a community detection algorithm, and looked at the extent to which mental maps overlapped with actual networks. he found that people are good at identifying the clusters they are involved in but are not good at identifying which of their social contacts act as `brokers' in the network. this finding has interesting implications - eg, since opportunities/new ideas tend to come from brokers and people find it difficult to identify brokers, then it follows that people do not know where to look for new ideas, right? ;) bernie also said that neurotics tend to have broken networks, while extroverts tend to have clustered networks. check bernie's publications here!

the student projects look very interesting. they include collaborative filtering, sentiment analysis, and community detection.

Filed under: Uncategorized No Comments
12Jul/112

UK SNA 2011

Posted by Daniele Quercia

few days ago, i attended the main social networks conference/gathering in UK.

there was an interesting discussion about the future of the elsevir journal "social networks". apparently, if you want to have an easy time getting in, you need to do research on 'methodology'. frankly, IMHO, this is the best thing they could do to kill the journal. alas, the journal's table of contents already reflect this decision. that is why i have rarely found interesting articles in this journal, while first monday and AJS are full of great contributions. don't get me wrong. i love methodological contributions to social networks - tom snijders and sinan aral are doing fantastic work in this area. i just think that methodological contributions are only a tiny part of a larger picture, a picture that hosts amazing work by, eg, duncan watts, danah boyd, and michael macy (all in US). instead, UK researchers in the area of social networks seem to be anchored to pretty "traditional research". at least, that was my impression based on the talks at UK SNA, but i will be very happy to be proven wrong ;) and there are notable exceptions in UK - eg, dunbar of oxford, bernie hogan of OII, and few others…

here are few notes taken during the talks.

cecile emery studied the relationship between big five personality traits and emerge of leaders. she considered not only leaders' personalities but also followers'. she found that leaders with high conscientiousness and extraversion tend to attract followers over time, and followers high in openness and conscientiousness tend to follow more. leader-follower pairs tend to be different on agreeableness and similar on openness.

agrita kiopa of georgia tech discussed a very interesting problem - how your friendship relations impact your work output. the main idea is that, to get something, you have to ask, so friendship becomes important also at work. they run a longitudinal national study of US academic scientists in 6 disciplines between 2006 and 2010. women are overepresented - i.e., 54% men, 46% women. friends are obtained by 6 name generators: role-based (collaborator, mentor) function-based (advice, discuss important issues), and close friends naming. 1600 egonetworks are collected as a result. so, for each person, there are 6 egonetworks. there is a considerable overlap among the 6 networks on average. full professors have more friends than assistant professors (control for tenure). the main results are that friendship has no effect in advice seeking but has effect on receive introductions and get reviews of, say, your papers. i hope she will devote a bit of future work to enemy (competitor) network. also, personality might be an interesting topic to study.

bernie hogan studied the correlates to social capital on Facebook. he used a mixed-method survey methodology and downloaded Facebook ego networks. he then focused on the question of whether your social capital is related to your (objective) network structure or to the way you (subjectively) perceive your network. Very interesting work.

tore opsahl's talk revisited the idea that small-world nets are ubiquitous. by contrast, he found that "small-world networks are far from as abundant as previously thought"

6Jul/110

netsci 2011

Posted by Daniele Quercia

i attended netsci a couple of weeks ago. here is my stream of consciousness:

lada adamic talked about how info changes as it propagates through the blogsphere, and she effectively modelled this change  as a simple urn model. more on her upcoming ICSW paper.  her future research will go into how sentiment of memes changes/evolves (this topic has been recently covered by  jure leskovec).

former navy officer duncan watts presented few macro sociological lab experiments and field experiments that he and his colleagues run on social media sites. he showed how media sites such as Twitter, Facebook, and Mechanical Turk allow researchers to measure individual level behaviour and interactions on a massive scale in real time. the good news is that  there are already guides for running experiments on those platforms  (see, for example, the tutorial at icwsm by paolacci  and mason).  the experiments he mentioned are fully reported in his latest book "Everything is Obvious". The main idea behind the book could be summarised as follows:

our intuition for human behaviour is so rich: we can "explain" essentially anything we observe. in turn, we feel like we ought to be able to predict, manage, direct the behaviours of others. yet often when we try to do these things (in government, policy, business, marketing), we fail. that is because, paradoxically, our intuition for human behaviour may actually impede our understanding of it. perhaps a more scientific behaviour would help us. the book is about experiments whose goal was to understand human behaviour at a large scale.

olivia woolley meza of  max plank presented the results of a project that measured the impact of two events (ie, island vulcan ashes and 9/11) on flight fluxes. these fluxes were modelled as  a network and metrics of interest were computed on the network - for example, they computed network fragmentation (nodes remaining in the largest connected component), and network inflation (how distances in the network decay). this study provided few intuitive take-aways, including:

  • regions geographically closer to an attack are more affected
  • between-region distancing is driven by hubs

with a presentation titled "A universal model for mobility and migration patterns", filippo simini (supervised by  marta gonzalez) turned to the question of whether it is possible to predict the number of flying/public transportation commuters between two locations. law of gravitation for masses (also called gravity model) doesn't work so well with people, and that's why they proposed a  new migration model.  the main idea of this model is that  an individual looks for a better job outside his home country and, as such, accepts the job in the closest country that has benefits higher than his home country. each country has a benefit value that  is a composite measure based on income, working hours, and general employment conditions. one  take away was that population (& not distance) is the key predictor of mobility fluxes.

giovanni petti of imperial skilfully delivered a very interesting presentation about a project called freeflow whose partners include  UK unis in london, york, kent. in this project, they have collected  data from sensors placed under speed bumps that measure the number of cars that pass and the amount of time each car has spent on a bump. traffic data has been arranged in a graph and, to identify congested areas, they run a  community detection algorithm  on the graph. it turns out that london behaves like one large giant in terms of traffic flow because of  long-range spatial correlations.

tamás vicsek studied  pigeon flocks and, more generally, studied the roles according to which birds tend to fly with each other. the main finding is that each member of a bird flock takes a specific role in a hierarchy, and roles arranged also  change over time

marta gonzalez studied the mobility of people living in different of cities (including non-US ones) and found few regularities:  for example, she found that residents of well-off areas tend to travel nearby (maybe because their areas tend to have plenty of internal resources). another interesting point is that trip length  distributions at city scale are well described by a weibull distribution. Marta also tried to reconstruct temporal activity of people living in chicago using SVD (eigen-decomposition) -  temporal activity is reconstructed only using the first 21 eigenvalues (which she called eigenactivities), and such reconstruction is also predictive of people's social demographics. the main point of this modelling exercise was to show that techniques such as principal component analysis combined with k-means are great tools to detect clusters in human activity

Filed under: Conference, Social No Comments