random notes & thoughts
From the Sunday's workshops, I remember this paper "Dating Sites and the Split-complex Numbers" It uses split-complex numbers to represent dating preferences in an elegant way. It seems promising. I'd be great to connect this work on previous papers on trust and distrust and on structural balance theories... I also heard that two presentations were quite good: 1) Content, Connections, and Context 2) Joseph Konstan talk abt the different decision strategies ppl have in different contexts.
On Thursday, we run a workshop on mobile recommender systems. Francesco Calabrese of IBM Smart Cities gave an interesting invited talk about current projects on transportation systems. Then, we had a set of really good talks & one outdoor activity. What did I learn? Well, most of the existing mobile systems assume that the recommendation process unfolds in one single step - get restaurant recommendations & choose one of them. In reality, recommendations in the built environment should go beyond that. For example,
- To mimic humans, the task of recommending restaurants should at least return 3 different recommendations (or facets): closest restaurant, best restaurant, trade-off between the two.
- One should understand WHY people visit certain places. How did they make those decisions? Which criteria did they employ?
- Recommender systems need to tap into established findings in the area of urban studies. For example, in our RecSys paper "Ads & the City", we exploited the fact that people are boring - they generally do not travel very far - unless what they are looking for is not readily available where they are.
- Temporal patterns in recommender systems have not been widely studied. They have been studied on Web platforms only recently (and Neal Lathia has done great work on that!) and have been neglected in mobile platforms. That is why we had another paper in the conference titled "Spotting Trends: The Wisdom of the few"
- Finally, and more importantly, we need far more user studies of how these systems are ACTUALLY used! Recommendations do not matter much -the experience counts ;)
And this is just scratching the surface ;)
I remember only few things from the conference (the industry track was pretty good):
- Multiple Objective Optimization in Recommendation Systems (linkedin). Nice example of A/B testing
- Towards Personality-Based Personalization (Thore Graepel of Microsoft Research). Nice talk about how easy is to predict personal attributes of Facebook users based on their likes. if you are interested in personality and social media, you should check out our work on Facebook and Twitter (we can predict personality traits of twitter users upon only their number of followers, following, and listed counts)
- Building Industrial-scale Real-world Recommender Systems (Xavier Amatriain of Netflix). Brilliant (& fully packed) tutorial. Check this out for a summary.
- Controlled experiments at Microsoft Bing (very good work): i encourage you to read 2009 guide [pdf] ; 2012 kdd paper; slides of the talk.
- Pareto-efficient hybrization for multi-objective recommender systems (UFMG). Here the question is how to combine different types of algorithms (hybrization).
- User Effort vs. Accuracy in Rating-based Elicitation (PoliMI). What's the optimal number of users ratings for movie recommendations? It seems to be between 5 to 20.
- TasteWeights: A Visual Interactive Hybrid Recommender System (UCSB). Visualization platform for your social media stream
- Learning to rank optimizing MRR for recommendations. Very cool work. It taps into the less is more concept, which I'm a big fan of
- Thumbs up to real-world stuff: Beyond Lists: Studying the Effect of Different Recommendation Visualizations; Yokie - Explorations in Curated Real-Time Search & Discovery Using Twitter; A System for Twitter User List Curation; The Demonstration of the Reviewer’s Assistant; CubeThat: News Article Recommender (browser extension for Chrome displays recommended additional news stories related to the same topic as the current news story)
- Challenges in music recommendation (@plamere from @echonest). A couple of interesting insights: "Understanding the specifics of your domain is critical to building a good recommender"; and recommending down-tail is OK, while recommending up-tail (britney to one who likes tom waits) is risky. Might be offensive to one's music identity. So make your recommendations Hipster-Friendly ;)
Communications and Multimedia Security
University of Kent, Cantervury
Sep 3-4, 2012
Proceedings are LNCS - will give to CL library if people want to look up any
Basic conference is fairly good- lots of low level detailed work...mainly securty, but some systems stuff
On Saturday I came back from the annual meeting of the UK Social Networks Association (UKSNA). It was the perfect place to meet old friends and make new ones. The program is here . I presented our work on why Twitter is a social network, and on why people "unfriend" each other on Facebook (the latter was covered by the New Scientist today). Interesting presentations include those on Tom Snijeders' multilevel longitudinal analysis of social networks, Harrigan's work on tie formation on Twitter, Emery's on "shared Leadership", and Marcus's on the connected communities project (those are the only presentations I could listen to - i joined a bit later because of other commitments in nottingham)
I was on a panel chaired by Bernie Hogan. The panel was discussing how network studies in the social sciences could help us to investigate the use of "Social Media". Here Paola summarised her thoughts on what was said, concluding:
"...there is an unmet and unrecognised need for theories. We don't know how social influence works in online networks. How social media help to form social capital, and how comparable this is to the social capital we knew offline. Whether local or global network effects are prevalent. How online/offline multiplexity works."
It's interesting to see that all these topics have been recently covered by people on a hybrid ground - that is, by people who are at the cross road of sociology, physics, and computer science. Namely, NB Ellison's work on building social capital in Facebook, Sinan Aral's on social influence online, and JP Onnela's on multi-slice networks. Development in the field could go no faster - by those who are not "in the field". Brokerage at work? :)
I just came back from WebSci - the "Web Science" conference. Papers were sane, keynote were plenty and of very high quality, the organisation of the conference was amazing.
Papers were mostly quantitative. One critical thinker unfairly criticized these papers to be Twitterology, and proposed to fix just that with additional work, which, others argued, manages to do Twitterology and Latourology (after Bruno Latour) at the same time - a mixed approach so-to-speak :) Unfortunately, only few papers were qualitative. I hope we will see a more balanced programme in the future - so, CHI & CSCW folks, please consider submitting your work ;) Well-grounded qualitative work is badly needed in this young community. Tip: I would use grounded theory...
quite surprisingly excellent - lots of interesting mixes of tech-aware media folks and media-aware tech geeks etc
some thoughts - evidence resistent theories about brands abound
politics is branded just like businesses - belief/trust failures are common in both
just being big eventually makes you "evil"
viral marketting doesn't (hardly ever) work...
page rank is ethical - auctioning advertises on search, social etc isn't.
crowdsources business models (financial and cellphone data plans) is cunning...but eventually they will revert to type...
I suppose there will be a formal output of the workshop...
social media folks have better slidepacks than tech/geek people:(
I can't really write a lot about it as I was only on the last session which was a celebration of the 40 years that Working Group 6 has been doing networking - quite impressive - I really didn't realize how joined up all the early thinking was - there was a great set of keynotes, especially by Dina Katabi (preview f some of her SIGCOMM 2012 work) and Pablo Rodriguez (some really insightful observations on extracting happiness from location data in cell phone logs- yes, privacy controversy was discussed and some of the excessive political power wielded by the big Cloud Gangs, surprisingly in a very impassioned talk by Louis Pouzin (the guy who invented datagrams, no less!). Good time in Prague, city of Robots and Golems:)
I´m at this 3rd e-energy conference in Madrid
Won´t see all the sessions, but some standout papers in session 2 including DTNs for dealing with intermittent energy (from UMass) and a solid comparison of VOIP v. PSTN energy efficiency (VOIP sucks, but you can do lots of smart enerfgy proportional work and duty cyling and eventualyl get it slightly better than an old PBX:)
plus a very nice paper from Wisconsin (Paul Barford´s group) on modeling power grid from 5 min spot price market for whole area ' LMP landscapes over time ' v. good stuff...
Standout session for me was Session 5 where more Waterloo work on why peak load pricing in canada is still too cheap to make energy storage worthwhile (paper 1 in this session= and teletraffic models of power (paper 4 in session 5)
Lots of other solid work on protocols (TCP over energy/rate proportional links, IP over lambda switches with duty cycling etc= and on optimisation...couple of ok game theory papers...mainstream stuff
Talked to lots of folks here from Waterloo (see ISS4e lab there!):http://blizzard.cs.uwaterloo.ca/iss4e/
Also, there was a chat from EU folks aboiut funding in this area, which is plentiful in next calls in summer (50MEu)
My keynote went down ok - some good questions about managing user demand for CDN content...
Other info from here:-