Archive for February, 2008

Filtering By Trust

Thursday, February 28th, 2008

A lot of the posts on this blog discuss the idea of computers using trust in different scenarios. Since starting as a research student, and reading a lot of good papers about trust, I still find the idea slightly strange: is trust something that can actually be quantified? Are these computers actually reasoning about trust? (what is trust?)

Trust, however, seems to have a place in our research. It gives us a metaphor to describe how interactions between “strangers” should happen, a platform to build applications that feed off these ideas, and a language to describe and analyse the emergent properties that we observe in these systems. So, even if a trust relationship may be a concept that cannot be boiled down to a deterministic protocol, it gives researchers a model of the world they are exploring.

I decided to apply this metaphor to collaborative filtering, an algorithm that aims to generate recommendations by assigning neighbours to each user. Usually this assignment is done by computed similarity, which has its downfalls: how similar are two people who have never rated items in common? What is the best way of measuring similarity? Applying the metaphor of trust, instead, aims at capturing the value that each neighbour gives to a user’s recommendations over time, and value is not only derived from agreement- but also from the ability a neighbour has to provide information. While similarity-based algorithms use neighbour ratings, interactions based on trust acknowledge that information received from others is subject to interpretation, as there may be a strong semantic distance between the way two users apply ratings to items.

Trust is a richer concept than similarity by favouring neighbours who can give you information about the items you seek and offering a means of learning to interpret the information received from them. Evaluating this technique improves on the basic similarity-driven algorithms, both in terms of accuracy and coverage: modeling the problem as an instance of a trust-management scenario seems to offer an escape from traditional pitfalls of similarity-based approaches. I leave all the details to the paper, “Trust-based Collaborative Filtering,” which is due to appear in IFIPTM 2008 (thanks to Daniele for his great comments and feedback).

Information Overload

Tuesday, February 26th, 2008

Many of us are researching ways of reducing information overload. The next The Economist Oxford-style debate revolves around information overload. The proposition: “This house believes that if the promise of technology is to simplify our lives, it is failing.” The pro speaker: Richard Szafranski (Toffler Associates). The con: John Maeda (MIT). (more…)

Workshop on Trust in Mobile Environments

Friday, February 15th, 2008

Following Daniele’s previous post on workshops at iTrust, another workshop is doing its own round of advertisement: the iTrust Workshop on Trust in Mobile Environments. Abstracts are due the 28th of March. Here is a short description:

Trust is a vital issue in mobile computing if applications are to support interactions which will carry data of any significance. Consider, for instance, exploring a market place: which vendors should one prefer, and why; how can a user establish the provenance of an item, etc. Various trust models have been developed in recent years to enable the construction of trust-aware applications. However, it is still not clear how robust these models are, and against what types of attacks; how accurate they are in capturing human characteristics and dynamics of trust; how suitable they are to the mobile setting. Mobility brings in orthogonal complexities to the problem of trust management: for example, the transient relationships with the environment and other users calls for an investigation of the dependency between trust and context; the lack of a clear shared control authority makes it difficult to verify identities, and to follow-up problems later; the limited network capability and ad-hoc connectivity require the investigation of novel protocols for content sharing and dissemination, and so on.

Much Ado About Six-degree (of the Few!) on the Road to Bin Laden

Sunday, February 10th, 2008

In social networks, after the nagging question of whether some of us are special (influential), the one about whether each of us is separated by only six individuals comes a pretty close second.

For the former, researchers’ answer was a qualified “yes” – changing only recently to be a qualified “no”. Influential individuals who start trends in society do NOT exist. On that, the relationship between social science circles and computer science circles seems peaceful enough. Social scientists have been publishing their findings upon which computer scientists are reverently adjusting their systems.

But mention the latter question, and the atmosphere changes. “Mon Dieu! Statistical Bullshit!” cry social scientists, while computer scientists get hotter around the collar of their Google shirts, look fretfully away, and studiously ignore the fuss. Or, to put it another way, social scientists

  • are coming round to the idea that there is simply not enough evidence that each of us is separated by six individuals. If we’re only six people away from bin Laden, why hasn’t he been tracked down and captured? “The answer, as Kleinfeld discovered, is complicated. It involves the misleading reporting of statistical data, the seductive power of a pleasing idea, and the vagaries of human behavior”. The truth seems to be that only a small fraction (roughly 29%) of people is separated from the rest of the world by only six degrees.
  • are saying (in the shape of Watts): “the question is not just whether we are closely connected, but how we navigate those connections—and whether we choose to do so at all. People can find these paths as long as they’re motivated to do so and able to motivate people to help them”.

Even so, computer scientists are anchored firmly to the idea of Universal Six-degree of Separation, pontificate on it in conferences, and seem to suffer from confirmation bias.

Two great workshops at iTrust

Friday, February 8th, 2008

1) Security and Trust Management (STM). Papers by April 2nd.
The intersection of security and the real world has prompted research in trust management. This research should ideally translate into proposals of solutions to traditional security issues. But, more often than not, it’s all proposals and few solutions. That is why STM focuses on how trust management may practically solve security issues and, in so doing, how it may enable new applications (eg, reputation, recommendation, collaboration in P2P or mobile nets). The call covers a wide range of topics.

2) Combining Context with Trust, Security, and Privacy (CAT). Paper abstracts by March 28th.
A research field might claim to have entered mainstream status only after it has been accepted by established conferences. Context-awareness and trust management have had that honour, but they have had it separately. We know by now how to design context-aware systems and trust management systems, but how to integrate the two is still the province of unexplored territory. That is why CAT will feature intrepid researchers who will stop us from:

  • sitting down in utter apathy towards the issue of trust being context-dependent – if (context=category of trust), as “rock music” is in “I trust you for recommending rock music”.
  • passing over exciting percom applications – if (context=space of interaction) as “my company premises” is in “my PDA is trusted for accessing confidential documents only within my company premises”.

Last year, CAT was terrific – I still remember the informing talks by Maddy, Tyrone and Linda. This year, it is likely to be even better. That is because CAT is like Math – one does context plus trust, and then multiplies by many researchers to equal stimulating discussion ;-)

Sell your stuff and shut up (unless you have good things to say)

Wednesday, February 6th, 2008

An article on the BBC is reporting that from May, eBay sellers will not be able to leave negative (or neutral!) feedback for buyers, since this  is “slowing down trade.” At the same time, they have not banned buyers from leaving negative feedback for sellers.

How word-of-mouth works

Monday, February 4th, 2008

The article Is the Tipping Point Toast? changes the way we think rumors (ideas) spread in social networks.

It works if one reaches the tipping point
“In a given process or system, some people (a tiny number of superinfluential types) matter more than others.” In modern marketing, this idea is the very premise of viral and word-of-mouth campaigns: Reach those rare, all-powerful folks, and you’ll reach everyone else through them, basically for free.

How to reach the tipping point (old way)
In the past few years, Watts (Columbia University & Yahoo) has challenged the whole Influentials thesis. He has analyzed email patterns and found that highly connected people are not, in fact, crucial social hubs. He has written computer models of rumor spreading and found that your average slob is just as likely as a well-connected person to start a huge new trend. And last year, Watts demonstrated that even the breakout success of a hot new pop band might be nearly random. Any attempt to engineer success through Influentials, he argues, is almost certainly doomed to failure.

How to reach the tipping point (new way)
“If society is ready to embrace a trend, almost anyone can start one–and if it isn’t, then almost no one can,” Watts concludes. To succeed with a new product, it’s less a matter of finding the perfect hipster to infect and more a matter of gauging the public’s mood. Sure, there’ll always be a first mover in a trend. But since she generally stumbles into that role by chance, she is, in Watts’s terminology, an “accidental Influential.”

Perhaps the problem with viral marketing is that the disease metaphor is misleading. Watts thinks trends are more like forest fires: There are thousands a year, but only a few become roaring monsters. That’s because in those rare situations, the landscape was ripe: sparse rain, dry woods, badly equipped fire departments. If these conditions exist, any old match will do. “And nobody,” Watts says wryly, “will go around talking about the exceptional properties of the spark that started the fire.”

Cascades require word-of-mouth effects, so you need to build a six-degrees effect into an ad campaign; but since you can never know which person is going to spark the fire, you should aim the ad at as broad a market as possible–and not waste money chasing “important” people.

Street Team ’08 – Citizen Journalists with Mobiles

Monday, February 4th, 2008

A group of 51 local reporters — one from each state and Washington, D.C. — will follow the 2008 elections and deliver weekly multimedia reports tailored for mobile devices.

Using short-form videos, blogs, animation, photos and podcasts, the reports will be streamed live all day on MTVNews.com and ChooseOrLose.com, and throughout the day, MTV will break into regularly scheduled programming to showcase news featurettes and live reports from our Street Teamers (whom you can meet here).

Social Graph API by Google

Monday, February 4th, 2008

The new Social Graph API “makes information about the public connections between people on the Web easily available and useful. You can make it easy for users to bring their existing social connections into a new website and as a result, users will spend less time rebuilding their social networks and more time giving your app the love it deserves”.

API docs & Google’s post.

Conferences and Festivals you should be aware of

Friday, February 1st, 2008

Ian Forrester’s post: “So there are a stupid amount of conferences and festivals in the UK this half of 2008…