<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type" />
<title></title>
</head>
<body>
<p style="margin: 0px;"><span> </span></p>
<div style="margin: 5px 0px 5px 0px;">
On September 6, 2011 at 12:03 PM Ivan Cukic <ivan.cukic@kde.org> wrote:<br />
<br />
><br />
> I've pushed the new contour daemon which unlike it was advertised doesn't<br />
> break everything. The dataengine is patched to work with the new daemon<br />
> and currently it shows some recommendations (at least on the desktop -<br />
> will test on exo when the image is available).
</div>
<p style="margin: 0px;"> </p>
<p>very nice!</p>
<p>we have created a new MeeGo image which you can use for testing, you can <span style="line-height: 21px;">download the latest MeeGo Plasma Active image from:</span></p>
<div class="plainTextContent" style="font-family: monospace; line-height: 1.5em;">
<a href="http://share.basyskom.com/contour/Deployment/latest-meego-plasma-contour-in-progress.html" target="_blank">http://share.basyskom.com/contour/Deployment/latest-meego-plasma-contour-in-progress.html</a>
</div>
<p>Cheers,</p>
<p>Karlheinz.</p>
<p> </p>
<p> </p>
<div style="margin: 5px 0px 5px 0px;">
> > There are currently two recommendation engines:<br />
> - the links (scripted) shows a few links (meant for testing purposes only)<br />
> - the mediapause - currently works with bangarang - while the player is<br />
> running, it shows the option to pause it. Completely useless, but since<br />
> there is no light sensor on the desktop, I wasn't able to do it like we<br />
> talked about. Will do it once this gets on exo.<br />
>
</div>
<div style="margin: 5px 0px 5px 0px;">
> The scored documents engine will come soon :) after that, we are on the<br />
> same level as the old contour daemon (ok, not the same, a few ft above)<br />
><br />
> ** Scoring the recommendation engines<br />
><br />
> I'm currently working on making the recommendation engines scored when the<br />
> user chooses an action.<br />
><br />
> The method is as follows:<br />
><br />
> 1 - the chosen engine's score is increased by a constant called $learning$<br />
> 2 - the list of recommendations is iterated - for each do:<br />
> - 2.1 - decrease the score of the engine by $learning * exp(-x)$<br />
> where x is the order of the recommendation in question,<br />
> starting with 1<br />
><br />
> How does this work?<br />
><br />
> The first thing is that what we are trying to keep the sum of scores after<br />
> the change the same as before. It doesn't really work, but it is close<br />
> enough since integral of exp(-x) from 0 to Inf is 1 - we don't have 1, but<br />
> something lower since in our case, we have a discrete sum of a finite<br />
> number of elements - exp(-1) + exp(-2) + ... + exp(-number_of_recomms).<br />
><br />
> Effects<br />
><br />
> - If an engine is too confident and gives a high score to its result, but<br />
> that result was not chosen, it will be punished more than other results<br />
> due to the exp(-x) part.<br />
> - the more recommendations an engine gives, the higher punishment it will<br />
> receive in the case none of its recomms are selected.<br />
> - if the engine behaves like in the previous example, but one of its<br />
> recomms is selected, it will have an increased score, but not as increased<br />
> as the engine that made only one recomm. This is due to the fact that<br />
> although it will get a reward for the good suggestion, it will get<br />
> punished for the bad ones as well.<br />
><br />
> I've been testing this on the two engines we have, and it is quite fun :)<br />
><br />
><br />
> Cheerio,<br />
> Ivan<br />
><br />
> p.s. I friggin' love the exp function :)<br />
><br />
><br />
> --<br />
> So remember when you're feeling very small and insecure<br />
> How amazingly unlikely is your birth<br />
> And pray that there's intelligent life somewhere up in space<br />
> Because there's bugger all down here on earth.<br />
> -- Monty Python<br />
><br />
> _______________________________________________<br />
> Active mailing list<br />
> Active@kde.org<br />
> https://mail.kde.org/mailman/listinfo/active<br />
>
</div>
<p style="margin: 0px;"> </p>
</body>
</html>