h a l f b a k e r yVeni, vedi, fish velocipede
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
The Problem: I am disorganised and have vast quantities of files, emails, documents etc that lie around in heaps on my computer's desktop. I do have a hierarchy of folders into which I sometimes sort my stuff. However it is usually too much trouble. More often than not I find that a file should be put
in more than one folder necessitating the creation of Shortcuts - giving me yet another reason not to bother.
The Solution?: Email spam filters use Bayesian classifiers to sort the mails you want to read from those you would rather not. My idea is to put the same technology into file managers. Just as spam filters can learn to classify mail into spam and non-spam, so file managers should be able to learn how to classify files into categories and sub-categories.
If I put a file in a particular folder the smart file manager is only allowed to keep it in that folder or move it to a descendant folder. This restriction is a learning opportunity for it to update its classifiers using the newly inserted file. In other words it adjusts its classifiers so that if the file had been inserted at the root of the file system then it would have been moved down to the folder into which it was actually inserted.
If, when browsing a directory, I see a misclassification I can easily correct it by cutting and pasting the file. The smart file manager again learns from this and readjusts its classifiers accordingly, improving its future classification capability.
After a bit of this kind of training I should be able to just drop the files at the root of the directory tree and the smart file manager will automatically classify them and place them in the correct directories. If it believes they should be placed into more than one directory then it will automatically generate shortcuts within those directories.
[Enhancement to idea based on [Freefall]'s comments]
1. I like the general idea that the system should just look like part of the filesystem and therefore be available to other applications.
2. As [Freefall] points out, this would cause problems as, under the current proposal, a file saved in one place might end up somewhere else. Although it must be in a descendent folder, I take the point that this would be confusing and inconvienient.
Therefore I would like to modify my idea slightly so that:
1. Files alway stay where you put them.
2. The system, instead of moving files to sub-folders just inserts shortcuts (links) in them.
The consequence of this is that the user has the option of looking for the file in the directory in which it was placed OR by looking in the most appropriate sub-folders.
and the 2nd
http://www.halfbake...dea/www.aicom.co.uk [engineer1, Oct 04 2004]
[link]
|
|
I really like this idea as I suffer from the same problem. This seeems like it could be baked, but one thing I was wondering about is how it would process non-text documents. Clearly word-processing, html, text, etc could be read, but images? music? I guess the file name could be used at least. I'd like to have a program that could do this well. |
|
|
Microsoft has it on the way, folks... stay tuned. |
|
|
I'm hip. I also am a relative bonehead when it comes to filing. Between computer-hopping and forgetting that every application I have seems to want to save files in its own unique path, I've lost it more times than I can count. |
|
|
You think your idea will help? + |
|
|
[matteus] Good point about the different file types. The way I see it working is that there would be a variety of plugin classifiers for different categories of things. For example a simple image classifier could work using image colours as classification attributes. Of course the user would not have to bother with this. There would be a load of classifiers bundled with the program. |
|
|
I avoid this by putting everything in a single directory. That way I always know where it is. Sure, it takes me half an hour to scroll through all my files and find what I'm looking for, but I always know what directory it's in. /end sarcasm |
|
|
Interesting concept, but without user input, this could get really ugly. Where did that file go? I don't know...the OS put it where it thought it should be, and now it's lost. Give me a window to confirm the placement (Saving file to "foldername\filename.xyz". OK or Select directory) and I'll bun it. |
|
|
[Freefall] I was thinking of this behaviour just belonging to a particular application (i.e. the smart file manager) rather than occurring throughout the entire OS. A person using that application would be expecting that behaviour. I agree, it certainly would be a good idea for the application to indicate where it had placed the files. But you always know that it is in the folder or in a subfolder of where you inserted it. For single files you could have a dialog box pop up. But I would want to dump a load of files in at a time. |
|
|
the [zigness] is right. This is not publicly baked yet, but Microsoft are working on something for future release that will do this. I believe the proposal is that files can be dropped into a repository and the OS categorises and files them for you. Anyhows, can't possibly divulge how I know this. |
|
|
Just made a slight modification to the proposal based on your comments. |
|
|
I would also be very interested to hear any more information about Microsoft's work-in-progress. Obviously I do not wish to exploit the idea commercially but I was thinking of starting an Open Source project to implement this. Does anyone know if they have an IPR on this? |
|
|
Erm... Someone's already done it!
-Ask for iStream info from Aicom Ltd (www.aicom.co.uk) - It's a rules based filing system... where _ANY_ filetype can have a bunch of rules defined against them... Rules can also fire-off other rules and can be nested too... It's very neat! - See also www.content-stream.com |
|
|
//Anyhows, can't possibly divulge how I know this.// Me either. |
|
|
Isn't this just some kind of fancy database?
It doesn't matter where the file is stored.. just how to find it? |
|
|
[Ling], not really. There's more to it in that heuristics are applied to create and manage the indexing. |
|
|
// I can't possibly divulge how I know this either! |
|
| |