Artificial Intelligence Depot
Visiting guest. Why not sign in?
News, knowledge and discussion for the AI enthusiast.
FEATURES COMMUNITY KNOWLEDGE SEARCH  
Flesh and Machines: How Robots Will Change Us
Antitechnologists may shudder at the story line, but readers interested in the astounding possibilities of the digital age will be fascinated by Brooks's vision of what is and what will be.
More information at Amazon US UK
Adaptive Websites
Describe them
 
Adaptive Websites

Hi there,

Im thinking of doing my university disatation on Adaptive Websites, but to me all they are are sites that are based on results generated when users hit certain content. Is this correct, could you give me a more detailed insight in to what they are and how they work?

Cheers,
Mike.

14 posts.
Monday 24 June, 09:47
Reply
Site Self-Optimisation

Sorry for the delay, flyingbuddha, I hope its not too late!

I've never heard of any adaptive sites in practice, but it'd be interesting to see one.

I'd consider the idea like this; to optimise the layout (set of actions controlling appearance) based on the user's trends (a feedback function that determines how good the layout is).

By mapping the problem onto a action/fitness, you can use standard AI techniques such as reinforcement learning. You'll learn the best actions that maximise the reward in the long term!

Does that make sense?

934 posts.
Tuesday 23 July, 12:44
Reply
Sorta similar

Hi there,
Thanks for the reply, no it isen't too late, Ive gone ahead with the idea and any input is useful. I have already started getting the basics behind a temporary site, its one that changes focus on elements that have higher views, and admin options on organising a site by various choices.
I think that is kinda what you were saying, but yours sounds more like the actual architecture of the layout, mine just moves better selling products to a designated spot so they are easier to see, etc.
Mike

14 posts.
Tuesday 23 July, 14:02
Reply
Optimising Clicks

Yes, I thought you meant organising the structure!

There are a few sites that optimise their article positions already. Check PHP Nuke. Basically, the most popular story is in the menu, and they have a top 10 X which showcases the best of everything.

That sounds like the sort of thing you want, but on a larger scale -- i.e., nearly everywhere in the site. Q-learning can do that nicely... you assign a Q value to each link, and increase it when it gets reward (== a click). Then sort by slightly randomised values of Q to display, in order to find the optimal policy. That's simple reinforcement learning. Check it out ;)

Keep in mind that pre-organised links are sometimes better. Constant layout, or layout in a particular order are sometimes very important.

Good luck!

934 posts.
Wednesday 24 July, 04:27
Reply
User Modeling and Adaptive Hypermedia

Hi Mike,

if you haven't already done so, you should take a look at the
User Modeling conferences (info and proceedings on-line at
www.um.org) and the Adaptive Hypermedia conference
(sirius.lcc.uma.es/AH2002/). In addition, Peter Brusilovsky's
site (www2.sis.pitt.edu/~peterb/) contains a lot of
interesting information about adaptive hypermedia in general.

I created a mildly adaptive website for the UM2001 conference.
It's still on-line at www.dfki.de/um2001, altough it's no longer
maintained. I used Bayesian Networks and Naive Bayes Classifiers
to estimate a user's interest in certain topics given evidences
extracted from her browsing behavior.

Regards,
Eric

1 posts.
Sunday 28 July, 08:27
Reply
Nice!

Wow,
Thanks, that information is really useful, I shall read when I have time.
Mike

14 posts.
Sunday 28 July, 16:31
Reply