BayesScheduler

[jimh]

I think that services like TIVO use statistics gleaned from its users as the basis for what shows it recommends/suggests you might like.

For example, let's say that TIVO gathers data that shows that 58% of people who regularly record "The Simpsons" also record "Scrubs", but only 6% of "Simpsons" fans also record "Will & Grace".  As a result, TIVO will suggest to you, if you schedule a season of the Simpsons, that you might also like "Scrubs".

Of course, in order for this to work on the GBPVR-level, GBPVR users will need to be willing to allow GBPVR or a stand-alone utility to share recording schedule data with a centralized database (anonymously, of course).

Then, GBPVR or the external utility can query the database for recommendations based on statistics collected in the master centralized database.

Programmatically, this probably is not that huge a task - far more challenging will be addressing controversial privacy issues, but those can be mitigated by allowing people to opt out of the service, something I don't think you can do with TIVO.

[jrockintuitive]

My curiosity is peaked...
Would there be weighted values? Maybe vote on a show -3 to 3 to get better results. Or something like that.

Correct me if I'm wrong, but the combination of results would probably not bear any fruit.
Meaning I watch Stargate SG-1 and JAG. And if Pegasus watches SG-1, JAG, and Battlestar Galatica (a great show), then the system would recommend BG.

[jrockintuitive]

jimh,

I can handle the messaging (anonymously) so there is no need to have a central server. We can use wiki for everyone to download a master list and some p-to-p, e-mail or whatever to build the list. We can even use the IRC that someone setup and transfer rating messages there.

I would need someone to take the lead on the logic to make recommendations.

What happened to the screen scrape idea with one of those recommendation sites?

[KingArgyle]

Uh... I think Jorm and I both got busy.

I'll have to dig up the info on that Website that allowed you to vote and then it would recommend shows for you based off your votes.

Jorm can you remember what the site was? I can probably find it again.

[Bigted]

Fantastic idea PegasusWar,
However, and I am not against Java in any way, is is possible to
program this using .net or whatever....?

I do not really want to have to maintain a JVM on top of
.net + everything else!

Am I being silly here of is it better to align software plugins to
the platform that you are writing for?
I could see this functionality being more closely integrated into
GBPVR over time and if so there may be issues around integrating .net vs java.

No holy wars please!!!!

[jorm]

King,

I do not remember the website. I did add a screen scraper that gets recommendations from imdb.com. I do not remember if it is in fastfind, xrecord or both.

It is called recommendations probably xrecord. It shows all of the recommendations in fast find.

Jorm

[KingArgyle]

Jorm, I know it was a wishlist thread, so I'll do a few searches there. A quick we search turned up this pretty decent site, quick and painless to get signed into as well.

http://www.mybestbets.tv/
It'll send you a daily, weekly, or monthly email recommending shows. You customize it by visiting the site, and giving the show recommendations a thumbs up or thumbs down.

Update:...I knew I would find it.
RatingZone.com

[jrockintuitive]

Jorm,

The imdb feature is fantastic. I played around with it a little this morning and hit a few bugs. I will report them to you when I get a chance.

What would be real cool if there was a feature in XRecord that would take all of you pending recordings (just the groups) and make a recommendation of shows you should be watching.
Basically, xrecord will process each show individual and compile one complete list. The list will have the recommended show (if you are not already recording it) and the amount of time that show is recommened - based on all of your pending shows. The most recommended show would appear first and so on.
Whatcha think?

I still plan on testing the bayes scheduler, the idea is very interesting.

[PegasusWar]

Quote[/b] (tipstir @ Mar. 22 2005,11:15)]So does this bayes program work or not, I am like most dont' want to load Java ontop of .NET, as I do use Java on another PC which has it's own task and not thing to do with GBPVR.

I understand your concerns about java in an environment that uses .net.
But as I stated at the beginning of this thread: it is only an experimental tool.
The target was: "is it possible to use spam technologie (Bayes...) to help scheduling recordings"
For me who don't have any knowledge in .net (but is willing to try), was the first logical step to try it in java.
If some user tried it and find it useful I am willing to try to port it in .net or give someone the source code to port it. But at this point I see no necessity to do it...

the java runtime env. is a small program that is not so integrated in the os as .net. If you surf on websites with java applets with firefox it is likely that you have it already on your system.

[KingArgyle]

Well, I personally have both .NET and Java on the same system with no problems, runtime is pretty much the same really. Anyway, if you did want to convert the application into a C# application, Microsoft does provide a migration tool that will do a lot of the work for you.

http://msdn.microsoft.com/vstudio/downloads/tools/jlca/

Since you know Java, learning C# is not going to be a problem, the languages are virtually identical with a few minor differences. I work a lot in Java and picking up C# was fairly easy.

No need to spend money on Visual Studio either, just get SharpDevelop or Borland C# Builder.