Archive for September, 2010

Ubicomp Day 2

Tuesday, September 28th, 2010

Here’s my pick for day 2: “Hapori: Context-based Local Search for Mobile Phones using Community Behavioral Modeling and Similarity“, from Dartmouth/Microsoft. This is a very neat example of context-aware recommender system, where POIs are being recommended to people

based on a rich variety of information, encompassing temporal information (day of the week, time of the day, etc), weather, and the activity of the user. As the authors summarise: “The goal of Hapori is to meet the diverse needs of different people (such as the teenager and senior) taking into account their context, behavioral profile and behavioral similarities with others in the broader community of local search users“.

Web Science: A Royal Society Meeting (day 1)

Tuesday, September 28th, 2010

notes on the first day of the royal society event web science. featured: lazlo barabasi, lord may, jon kleinberg, and david karger ;-)


Ubicomp Day 1

Tuesday, September 28th, 2010

My pick from the first day at Ubicomp: “Rethinking Location Sharing: Exploring the Implications of Social-Driven vs. Purpose-Driven Location Sharing” (from CMU). Key idea: the rationale behind sharing location has a big impact on people willingness to share and what they share. The main distinction is between sharing for a need (purpose sharing, for example, to get phone calls redirected to where I am – like in Active Badge), versus sharing because I want (social sharing, for example to stay more connected with my friends). Their study reveals that location sharing varies a lot between the two cases (need vs want), with different labels being picked by people to describe places, and with different privacy concerns being attached to them. Another more

privacy-focused paper from the same people was also interesting (“Empirical Models of Privacy in Location Sharing“).

meeting on privacy at imperial college

Monday, September 27th, 2010

Last week, I gave a talk titled “Promoting Location Privacy … One lie at a time” at a workshop on privacy at Imperial. The slides of all

the talks are here

Ubiquitous Crowdsourcing

Sunday, September 26th, 2010
I’ve been attending the Ubiquitous Crowdsourcing workshop today @ Ubicomp. We were a small how much clomid to take group of 10 people only, but the quality of the talks has been really high, and the discussion that followed very intriguing. A nice blend of applications and theory was presented. What it emerged was a pretty much consistent picture of what the challenges in crowdsourcing are: for example, incentives, quality assurance, business models. However, what also emerged from the presentations was that solutions widely vary. During discussion, social scientist Thomas Erickson from IBM suggested a classification framework to characterise different crowdsourcing applications, which may help scientists understand what solutions fit best their specific problem. The framework is simple yet very useful, and distinguishes 2 orthogonal dimensions only, time and place: same time same place (e.g., audio-centric apps), different place same time (e.g., an enterprise language translation tool presented by the keynote speaker, Uyi Stewart, also from IBM), same place different time (e.g., Cyclopath, presented by Thomas himself), and different place different time (a la wikipedia).
The picture is not as simple as that though. Michael N. Huhns, from the University of South Carolina presented a case study where architects

built a new university campus WITHOUT roads, let people walk for a year around it, THEN built paved roads following the paths that people used the most. It never occurred to me that the simple act of “walking” could be seen as crowdsourcing, I always thought crowdsourcing required some intention, never mind how simple the task is. But here we go: un-intentional vs task-driven crowdsourcing!

Last thought: crowds vs communities. Lots of work goes on to create incentives to retain and sustain a crowd. At what point will the crowd (meaning a set of individuals working somehow competitively towards a task) becomes a community (where competition disappears and is transformed into cooperation)? And what are the consequences?

forgetting for recsys

Sunday, September 26th, 2010

Before the digital age, remembering was costly and hard, and the default for humans was to forget. Forgetting is a good thing for a society, not least because people are willing to engage (they don’t fear the recall of trivial past deeds) and take better decisions (forgetting allows human decision-making to generalise and abstract from individual experiences). In the digital age, the balance has been inverted: remembering is cheaper and easier than forgetting. I was thinking about possible technological solutions that make forgetting a tiny bit easier than remembering in the digital world. So my question is: what if you had a recommender system in which you could specify the expiration date for each of your ratings? What would you do? Recommender system for online shopping might be a case in point (“past online purchases” are used to recommend items you might like). Forgetting might be useful for users who order birthday presents in that

they accutane 10mg might opt for  short expiration dates for those  purchases,  so future  recommendations for them would not be influenced by their purchases for somebody else. Any other practical idea is welcome!!! ;-)

Random Picks from ECML/PKDD 2010

Friday, September 24th, 2010

I’m at the 2010 ECML/PKDD Conference in Barcelona, Spain; I’ll be presenting a paper (slides, paper) at the Workshop on Privacy and Security in Data Mining and Machine Learning (PSDML). Check out the real-time posts on twitter here.

Also, in a similar fashion to what Daniele did for CIKM 2010, here is a

completely arbitrary, title-biased pick of papers from the conference:

  • Demand Drive Tag Recommendation” by G.V. Menezes, J.M. Almeida, F. Belen, M.A. Goncalves, A. Lacerda, E.S. de Moura, G.L. Pappa, A. Veloso, N. Ziviani
  • Fast and Scalable Algorithms for Semi-Supervised Link Prediction on Static and Dynamic Graphs” by R. Raymond and H. Kashima
  • Collecting Traffic Forecasting” by M. Lippi, M. Bertini, P. Frasconi
  • Selecting Information Diffusion Models Over Social Networks for Behavioral Analysis” by K. Saito, M. Kimura, K. Ohara, H. Motoda
  • Finding Critical Nodes for Inhibiting Diffusion of Complex Contagions in Social Networks” by C.J. Kuhlman, V.S.A Kumar, M.V. Marathe, S.S. Ravi, D.J. Rosenkrantz
  • Analysis of Large Multi-Modal Social Networks: Patterns and a Generator” by N. Du, H. Wang, C. Faloutsos
  • Surprising Patterns for the Call Duration Distribution of Mobile Phone Users” by P.O.S. Vez de Melo, L. Akoglu, C. Faloutsos, A.A.F. Loureiro
  • Virus Propagation on Time-Varying Networks: Theory and Immunization Algorithms” by B.A. Prakash, H. Ton, N. Valler, M. Faloutsos, C. Faloutsos

The last four picks were, unsurprisingly, from the social-networks session.

Best Paper Awards:

  • Large Scale Image Annotation: Learning to Rank with Joint Word-Image Embeddings” by J. Weston, S. Bengio, N. Usunier
  • Accelerating Spectral Clustering with Partial Supervision” by D. Mavroeidis
  • A Geometric View of Conjugate Priors” by A. Agarwal and H. Daume III
  • A Game-Theoretic Framework to Identify Overlapping Communities in Social Networks” by W. Chen, Z. Liu, X. Sun, and Y. Wang

Workshops, with some picks:

  1. Analysis of Complex Networks
  2. Mining Ubiquitous and Social Environments
    • Towards Adjusting Mobile Devices to User’s Behaviours” F. Jungermann, K. Morik, N. Piatkowski, O. Spinczyk, M. Stolpe, P. Fricke
    • Discovering Trend-Based Clusters in Spatially Distributed Data Streams” by A. Ciampi, A. Appice, D. Malerba
  3. Supervised and Unsupervised Ensemble Methods and their Applications
  4. Preference Learning
    • Semantic-Based Destination Suggestion in Intelligent Tourism Information Systems” by M. Ceci, A. Appice, D. Malerba
  5. Handling Concept Drift in Adaptive Information Systems: Importance, Challenges, Solutions
  6. Third Generation Data Mining: Towards Service-Oriented Knowledge Discovery
  7. Privacy and Security Issues in Data Mining and Machine Learning
    • Content Based Filtering in On-line Social Networks” M. Vanetti, E. Binaghi, B. Carminati, M. Carullo, E. Ferrari
  8. Dynamic Networks and Knowledge Discovery
    • Stream Based Community Discovery via Relational Hypergraph Factorization on Evolving Networks” by C. Bockermann and F. Jungermann
    • Collaboration Based Social Tag Prediction in the Graph of Annotated Web Pages” by H. Rahmani, B. Nobakht, H. Blockeel
    • Automatic Sentiment Monitoring of Specific Topics in the Blogosphere” by F.S. Pimenta, D. Obradovi, R. Schirru, S. Baumann, A. Dengel
  9. Detection and Identification of Rare Audiovisual Cues
    • Anomaly Detection and Knowledge Transfer in Automatic Sports Video Annotation” by I. Almajai, F. Yan, T. de Campos, A. Khan, W. Christmas, D. Windridge, J. Kittler

CIKM 2010: my arbitrary pick

Friday, September 17th, 2010

The program of CIKM has just been released pdf

I randomly fished some papers. By randomly, I mean that I chose  papers based on whether a title/abstract suggests a research problem of our readers’ interest. Here you go.

  • Preserving Location and Absence Privacy in Geo-Social Networks
  • Inferring Gender of Movie Reviewers: Exploiting Writing generic viagra united states Style, Content and Metadata, Jahna Otterbacher
  • Pricing Guaranteed

    Contracts in Online Display Advertising

  • The Gist of Everything New: Personalized Top-k Processing over Web 2.0 Streams,
  • Web Search Solved? All Result Rankings the Same?
  • Travel Route Recommendation using Geotags in Photo Sharing Sites
  • Boosting Social Network Connectivity with Link Revival
  • Power in Unity: Forming Teams in Large-Scale Community Systems
  • You Are Where You Tweet: A Content-Based Approach to Geo-locating Twitter Users
  • Collaborative Future Event Recommendation
  • EUI: An Embedded Engine for Understanding User Intents from Mobile Devices
  • n WikiPop – Personalized Event Detection System Based on Wikipedia Page View Statistics
  • Evaluating, Combining and Generalizing Recommendations with Prerequisites
  • Classical Music for Rock Fans?: Novel Recommendations for Expanding User Interests (great problem. i hope to see more on this ;-))
  • Predicting Product Adoption in Large-Scale Social Networks