A Blog recording the life a Web Scientist
Network Science and Web Science 2012 Summary
The Network Science and Web Science 2012 were held back-to-back on the 18th – 24th June 2012 in North Western University, Chicago, USA. This was the first time these conference were ran together at the same time and location and due to their cross over in research topics, promised to be a great week of presentations, discussions, learning and networking. Both disciplines that these conferences represent are relatively new fields of research, thus a great breeding ground for novel and exciting research. The conference participants, presenters and keynote speakers all come from a variety of disciplines and research interests, ensuring a true interdisciplinary environment – this was reflected in the accepted papers for both Web Science and Network Science.
The following discussion will describe some of the highlights of both conferences, including a selection of the best papers, keynotes, and general outcomes of Network Science and Web Science 2012.
Network Science 2012
Network Science, a recently formed discipline draws upon a multidisciplinary field of researchers, including from Mathematicians. Physicists and Computer Scientists, but also draws upon those interested in Psychology and the Social Sciences.
The first two days of the conference was occupied with a number of Workshops, and also a Network Science School, which offered an jam packed 2-day introduction into the field of Network Science; although I didn’t attend this, I was told that it was an extremely useful and informative 2-day program, offering some great learning outcomes and also connections with those already researching in the field of Network Science.
Attending the ‘Languages and Network Science’ workshop on the first day was a great way to jump into the conference – focusing on the use of network science to help model a number of problems and research topics associated with language learning, processing and semantics.
The Workshop – a full day programme – kicked off with presentations regarding the modelling of phonological network structures, first examining how children learn and increase their corpus of words and then moving on towards examining the connectivity of words, how they are learnt and the semantic similarities that they share. The use of Network Science approaches to modelling these networks provided a test-bed for experiments with language networks, from modelling how certain words are learnt by children at different ages, to examining the mapping of semantics between words. Although this is not my current area of research, the morning session was very interesting and useful to attend, it demonstrated the broad range of research that can be classed at network science, However, I was expecting to see much more real-world empirical data within experiments, which, as this report will raise again, tended to be the typical methods used within the research presented.
The afternoon of the Languages and Network Science workshop soon became populated with research on social networking and collaborative sites such as Twitter and Wikipedia. Paolo Masucci gave an interesting paper on the semantic flow of language between different Wikipedia pages, using the Italian mafia as a great example of how semantics links between Wikipedia pages can be used to build up networks of connections. In addition to this, Xiaoju Zheng’s presentation on the use of Twitter hashtag dispersion was extremely relevant, providing some figures such as 98% of hashtags are words, and that well dispersed – spread amongst multiple user communities – are more likely to stay popular and not die. Words2Play.com was also demoed, which is a social machine which makes a game out of splitting up blended hashtags (i.e. twestival, twitter and festival).
Overall, the workshop offered a good start to NetSci12, setting the standard high for the next few days of presentations and discussions. Although the selection of presentations in the afternoon were more geared towards empirical-based research, what was noted though was the heavy focus towards theoretical modelling of single variable problems, something that I was not expecting or researched into before this workshop. Furthermore, another take away lesson from day one was the extent to which Network Science is being used in other disciplines, not only within the computational sciences, but physical and social sciences as well; and within each, using the theory and methods of network science to tackle their research problems.
The second day of Network Science consisted of a tutorial on the Sci2 – Science of Science – tool, a similar tool to Network Workbench, offering analysis and visualisation of large network datasets. The half day workshop provided a hands-on walkthrough of the tool, demonstrating its various features and capabilities – definitely a great way to quickly become efficient in the use of the software.
The official opening of NetSci2012 started off with two keynote speakers: Luis Amaral and Iain Couzin. Luis gave an insightful personal perspective on the field of Network Science and Iain provided an exciting presentation on his work looking at social networks in animals and collective behaviour; fascinating work, demonstrating that animal social networks are usually based and tracked by proximity, and the strength of interactions are based on the number of interactions that they have with others, thus the individual decisions and ‘opinions’ of the animal are closely linked to the local majority.
The afternoon session was split up into 5 parallel sessions which ranged from social networks to fundamental network properties. Attending the social network session, a range of interesting and relevant presentations were given, with highlights from Alex Rutherford’s study on mobilizing people fast – modelling the DARPA challenge experiment, reporting that distant or long range friends tends to be more active than those in close proximity. Looking at this session with my Web Science hat on, Sameet Sreenivasan’s research on tipping points of views/opinions within social networks was also very relevant, demonstrating the push needed to create a change in political positions of social network users – demonstrating that during the transitional period, there tends to be an intermediary state, which represents the indecisiveness of users.
Attending the two social networking sessions in the afternoon, the reoccurring theme, which I mentioned before was the use of theoretical models to examine different kinds of social phenomena, yet I felt some of them could have been made stronger with the use of more empirical data – I put this down to disciplinary differences though, despite this, the attended talks were insightful and demonstrated the application of Network Science.
The second day of NetSci12 began again with a selection of keynotes and invited speakers, one which was particularly interesting was Neil Johnson’s research on ultrafast financial transactions, examining how the speed at which transactions are produced is important for “global control”.
Reflecting on the Network Algorithms and Network Measures tracks in the afternoon session of day 2 (although some presentations were beyond the scope if my understanding!), a number of new novel approaches were given for dynamic analysis of networks, an area which I have been interested in for quite some time. The argument for dynamic analysis to understand the changes within a network was clear, and was also applied to help identify different communities and groups within networks – something well worth looking into!
The conference dinner was also a worthwhile trip, as it featured a after dinner keynote from Barabasi, discussing the evolution of Network Science and his current work within the medical sciences. Some really interesting ideas (and graphics) were shown, especially the application of new approaches in network science to help map out the pathways of human diseases.
The third and final day of NetSci was a morning filled with excellent keynotes by James Fowler and Lada Adamic; James’s talk looking at large scale networks and examining social influence and political mobilisation – with a really good take away message from it being: You should ask the question of why things happen, but you need to ask why things don’t happen! Lada’s talk – on social memes – was timely and had a lot of relevance to some of the work that I’ve been doing on influence and diffusion of messages within social networks. The talk really gave a good overview of the diffusion of memes within social networks, how they change, adapt and become something completely different from the original along the diffusion path – notably a growing area of research. The morning session finished with a light-hearted talk by Michael Macy on “why do liberals drink lattes” – which examined political preferences in social networks.
Finishing the morning session was the signal for the end of NetSci2012 and the beginning of Web Science 2012, with an adjoining keynote given by John Kleinberg on Status and Evolution in online social networks.
Web Science 2012
The Opening Keynote of Web Science 2012 was given by Kleinberg, who talked about social status and feedback effects in online social networks, discussing two interesting theories, that of balance and status, which examined the role of status and power that individuals have within a network, and as Kleinberg discussed, those with high status tend to be close to other with high-status, yet, the rate at which you are respected is lowest amongst those at the same status as you. Furthermore, those with low power tend to act more as the coordinators within a network.
After the keynote, the first session was a presentation session on social network and friendships, in particular, Daniele Quercia et al. paper on Loosing “Friends” on Facebook was fascinating, demonstrating that the lack of common ‘metadata’ amongst users was strongly correlated to the unfriending of individuals, with age playing a major factor.
The panel session which followed this – Social computing and collective intelligence – saw some really diverse and interesting research, from examining recommendations on (food) ingredients Websites to the use of social media platforms to enhance daily deal Web services. After listening to the panel discuss their work, the underlying theme of these presentations seemed to fit in with the concept of the Social Machine, how they can be designed (which is part of a social and technical process), and how they become accepted and used by the masses. An interesting thought is can existing social machines – Twitter, Wikipedia – be used as a platform of springboard for new machines, can they grow on-top of/side-by-side each other?
The Second day of Web Science opened with a keynote from Sonia Livingstone, who talked about the use and affordances of digital technological in and around the classroom. This was a much welcome talk to the Web Science community, as it stepped outside the usual boundaries of tech-heavy talks towards one that was much more about real-world experiences and in-the-field research. Following this, Clare Hooper presented a great opening presentation on the cross over between HCI and Web Science, and was an excellent piece of research asking (and providing answers) to some of the fundamental issues within the Web Science discipline –discussions that are well long over-due! Similar to the research that I have been working on, Clare pushed forward the idea of qualitative and quantitative methods within the intersect of HCI and Web Science, arguing their own individual strengths and suitability. Terhi Nurmikko – A fellow Web Science student from Southampton – also gave a brilliant presentation of the use of semantic Web technologies within Cuneiform Studies, offering a great solution to help the niche field with ways to make documenting and working with artefacts much easier, it was great to see the field of archaeology being represented at Web Science.
The afternoon keynote was given by Sinan Aral, who discussed measuring influence in social (media) networks –an area which I’ve been interested in for quite some time, especially with my current work with Edelman who is interested in developing tools to examine influence within social networks such as Twitter. The presentation gave a detailed overview of the difficulties of detecting and classifying influence within a social network, the problems with casual estimations of influence, drawing upon concepts such as Constructed Observational evaluation to discuss alternative ways to deal with detecting influence. As part of the talk Sinan discussed a recent Yahoo study which examined 27 million users and found that you are 16x more likely to adopt a technology if an individual has a friend of a friend that does so to. Statistics like this make me wonder how quickly users could migrate to a new social media service (like the Facebook vs. Google+ debate), and also adds to the argument of the Web being a temporally stabilized set of networks, only held together by their continuous support and activities.
Following the keynote, the second presentation panel of the day began kicking off with Chris Phethean, another fellow Southampton Web Scientist. His talk on the use of social media within charities looks to be a promising area of research, examining how social media can be used to help further and support a campaigns and audience base. After a short break, the next panel session began, with a great series of papers and discussions concerned with ‘democracy, policy and the Web’. Two standout papers were given: Kieron O’Hara’s Transparency, Open Data and trust in the Government, and Sabrine Saad et al. (presented by Stéphane Bazan) research on the Infowar in Syria. Some really great in-action research, highlighting the real dangers of the Web, the consequences of its use within political uprising (Syria Electronic Army).
The end of day 2 was closed by a keynote from Danah Boyd who discussed the ever growing threat of privacy on the Web, drawing upon her extensive ethnographic research to show how data privacy is changing, how individuals are dealing with it currently (like teenagers hiding meaning rather than content), and how in future we may need to implement some formal system or policy to protect society. A very important point was made during the keynote: data is like DNA, you can’t share your own without sharing someone else’s. The talk also sparked the debate to what was ethical within the world of ‘Big Data’ research, just because the data is out there, does it mean that it should be used? Will the data that is currently available now be (ethically) usable in the future? It is these kinds of (Web Science) questions that really need to be explored further.
The final day of Web Science 2012 began with an Industry panel session, which then was followed by a diverse (yet technical) set of presentations on social network analysis. In particular Lars Backstrom et al. paper on “Four Degrees of Separation” was very interesting, which analysed the strength of connections and distance within the Facebook social networking site, revealing that in 2008, the average distance was 5.82, but in 2008, it reached 4.74 (4.32 in the US)! Great findings, but this really needs to be put into context, as the study is only considering Facebook as the source of evidence, Web Science needs to be careful not to make too many claims, as (as Lars said himself, the media tend to report things wrong, or entirely misinterpret it).
After lunch, the final presentation session on “Methods and Applications” began, with me being the last presentation to be given. The diversity of the research being presented were good, with Hans Akkermans’s presentation on how power laws occur, to Jérome Kunegis discussing fairness on the Web, introducing an alternative approach to the power law for understanding the long tail effect of the Web – the Gini coefficient. Following this, I presented my work on Mixing methods and Theory to Explore Web Activity, which focused on the use of using both qualitative and quantitative data to show that Web activity is a product of complex socio-technical networks that need to be understood not only from the online and offline perspective, but also at the micro and macro level, and whilst doing so, maintaining a perspective that isn’t deterministic. The presentation got some really great responses (and challenges), and the argument against the growing trend in Twitterology (coined by Professor Catherine Pope) was put forward. This definitely created a (well-needed) stir in the conference, sparking lots of good debate!
To end Web Science 2012, a well-known figure in the Computer Science (and Web Science) discipline closed the session – Luis Von Ahn, who developed the highly popular (and somewhat frustrating) ReCAPTCHA service. His keynote was both light hearted – ReCAPTCHA art and the church of Inglip – and thought provoking, not only revealing the new language translation/learning social machine DuoLingo, but also posing the question of “How do we coordinate 10 million to solve crime?” – A question which was directed at the use of social computation to solve the high crime rate in South American countries. Although this sparked a number important political, social and philosophical issues (which, without a doubt was a good thing), asking these kind of questions are the kind of things that need to be done to push Web Science in new directions – it is only with these kind of questions that we can start to unravel issues associated with it.
Both Network Science and Web Science 2012 were great – the speakers, the location, the organisers and the weather! The week provided many thought provoking debates, ideas, and new research areas/interests. The keynotes chosen were excellent, providing a well-balanced perspective of the current direction and challenges that these fields face. With Network Science 2013 located in Denmark and Web Science 2013 in Paris, it’s time to get writing again!