The death of traditional file names and directories
-
Jim Crafton wrote:
f-stop, aperture, dimensions, DPI
And that's helpful how? I can just see my wife thinking; where's that picture of the kids at the beach? I remember it used F/4 at ISO 100...
Well then you might search for pictures of kids at a beach. There's an article here: Automatic Linguistic Indexing of Pictures (ALIP) By Artificial Neural Network Approach[^] that talks about doing just that. Maybe a system might come with a default dictionary of tags, so a new file that got added would be "tagged" automatically.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
In 2009 why do we bother with directory names and traditional file names any more? What would be some alternative ways of organizing things? I was wondering about this because my sister-in-law, people I know at work, and even my parents often struggle with remembering what a file is called, giving poorly chosen names to files and then being unable to find them quickly, and so on. I'm sure others here have had to help family members, or maybe even co-workers find "misplaced" files. Yet one of the things I thought was cool about a Palm Pilot was that you never worried about files, or where things were stored. Why can't this be applied to a PC OS? Why not use a combination of things? Something like tagging to add descriptive bits, a UI that has alphabetic index, like a book. Given that NTFS has many of the features in place to do something like this (Alternate Data Streams would be a perfect place to store meta data, tags, etc) how hard would it be to write something that might accomplish this? What other things would be nice to have?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
here's how this idea went over in 2006: http://www.codeproject.com/Messages/1333841/Re-Making-Computers-Idiot-Proof.aspx[^]
-
In 2009 why do we bother with directory names and traditional file names any more? What would be some alternative ways of organizing things? I was wondering about this because my sister-in-law, people I know at work, and even my parents often struggle with remembering what a file is called, giving poorly chosen names to files and then being unable to find them quickly, and so on. I'm sure others here have had to help family members, or maybe even co-workers find "misplaced" files. Yet one of the things I thought was cool about a Palm Pilot was that you never worried about files, or where things were stored. Why can't this be applied to a PC OS? Why not use a combination of things? Something like tagging to add descriptive bits, a UI that has alphabetic index, like a book. Given that NTFS has many of the features in place to do something like this (Alternate Data Streams would be a perfect place to store meta data, tags, etc) how hard would it be to write something that might accomplish this? What other things would be nice to have?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
I suspect the web model of sites like flickr and YouTube are the best current examples of this. They make items easy to find through use of automatically identified metadata (dates and camera information, for example) and manually assigned values (tags). They do support some organizational facilities (sets and groups on flickr), but they are neither required nor sophisticated. The problem with this model is that it is indiscriminate. It's not vital to the end user to find an exact document. Anything that matches the search terms will do. This retrieval mechanism may work for some things (music is an obvious example), but not others (financial records). Many applications are self-organizing in this regard: e-mail and media players are easy examples. In those cases, individual 'documents' have a naturally assigned identity. E-mails have senders/receivers, dates, and subjects. Songs or movies have titles and artists. Material that the user authors himself is a lot more difficult, because you then rely on him to supply the identity directly, or at least sufficient hints to construct the identifying metadata. I have a feeling the late lamented WinFS feature was an attempt to deal with this issue, and they realized it was a harder problem than anyone thought.
Software Zen:
delete this; -
here's how this idea went over in 2006: http://www.codeproject.com/Messages/1333841/Re-Making-Computers-Idiot-Proof.aspx[^]
Yikes, not so well :(
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
Well then you might search for pictures of kids at a beach. There's an article here: Automatic Linguistic Indexing of Pictures (ALIP) By Artificial Neural Network Approach[^] that talks about doing just that. Maybe a system might come with a default dictionary of tags, so a new file that got added would be "tagged" automatically.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Or, you do what I do and a) copy all files from the camera into a folder with the date the copy was made (i.e. 2009_07_01) and b) rename the good pictures with the date, who is in them and where it's at. (Automatic index results would be interesting. I wonder how it would index the pictures of my kids at the beach [in San Diego a month ago] since the beach isn't in several of the pictures. Moreover, there were multiple beaches. Even then I'm not sure it's a help. To my wife, the two main beaches were "The Hotel Beach" and "The Other One." To me, they were "The Bahia Beach" and "Pacific Beach." To my kids they were "The Boring Beach" and "The one with the waves.")
modified on Wednesday, July 1, 2009 4:33 PM
-
I suspect the web model of sites like flickr and YouTube are the best current examples of this. They make items easy to find through use of automatically identified metadata (dates and camera information, for example) and manually assigned values (tags). They do support some organizational facilities (sets and groups on flickr), but they are neither required nor sophisticated. The problem with this model is that it is indiscriminate. It's not vital to the end user to find an exact document. Anything that matches the search terms will do. This retrieval mechanism may work for some things (music is an obvious example), but not others (financial records). Many applications are self-organizing in this regard: e-mail and media players are easy examples. In those cases, individual 'documents' have a naturally assigned identity. E-mails have senders/receivers, dates, and subjects. Songs or movies have titles and artists. Material that the user authors himself is a lot more difficult, because you then rely on him to supply the identity directly, or at least sufficient hints to construct the identifying metadata. I have a feeling the late lamented WinFS feature was an attempt to deal with this issue, and they realized it was a harder problem than anyone thought.
Software Zen:
delete this;Gary Wheeler wrote:
because you then rely on him to supply the identity directly
If you rely 100% on the user I'd agree, but with the processing power that we have, couldn't the OS examine the file and generate some initial meta data? For example I'm sure there are AI projects out there that can scan a text file and give you some clues to it's content. Is it a legal document? Is it a letter? Etc, etc. Ditto for pictures, there's a article here that uses some AI techniques to identify pictures by high level concepts, i.e. a picture of a bird or a cat.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
Or, you do what I do and a) copy all files from the camera into a folder with the date the copy was made (i.e. 2009_07_01) and b) rename the good pictures with the date, who is in them and where it's at. (Automatic index results would be interesting. I wonder how it would index the pictures of my kids at the beach [in San Diego a month ago] since the beach isn't in several of the pictures. Moreover, there were multiple beaches. Even then I'm not sure it's a help. To my wife, the two main beaches were "The Hotel Beach" and "The Other One." To me, they were "The Bahia Beach" and "Pacific Beach." To my kids they were "The Boring Beach" and "The one with the waves.")
modified on Wednesday, July 1, 2009 4:33 PM
Yeah, that's great, and it's what I do to, but lets face it, most people don't. And given that most people aren't going to do this, and have ZERO interest in even bothering, I'm curious if it's possible to get the OS to "help" out a little more here. It just seems to me that with all the R&D that's been done over the years, and with all the spare cores that people have on their computers doing very little work, something like this should be possible.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
In 2009 why do we bother with directory names and traditional file names any more? What would be some alternative ways of organizing things? I was wondering about this because my sister-in-law, people I know at work, and even my parents often struggle with remembering what a file is called, giving poorly chosen names to files and then being unable to find them quickly, and so on. I'm sure others here have had to help family members, or maybe even co-workers find "misplaced" files. Yet one of the things I thought was cool about a Palm Pilot was that you never worried about files, or where things were stored. Why can't this be applied to a PC OS? Why not use a combination of things? Something like tagging to add descriptive bits, a UI that has alphabetic index, like a book. Given that NTFS has many of the features in place to do something like this (Alternate Data Streams would be a perfect place to store meta data, tags, etc) how hard would it be to write something that might accomplish this? What other things would be nice to have?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Jim Crafton wrote:
UI that has alphabetic index, like a book.
That is essentially what a directory [structure] is. The files are just there in the disk, and all those tagged with a certain directory 'belong' to that directory.
-
Gary Wheeler wrote:
because you then rely on him to supply the identity directly
If you rely 100% on the user I'd agree, but with the processing power that we have, couldn't the OS examine the file and generate some initial meta data? For example I'm sure there are AI projects out there that can scan a text file and give you some clues to it's content. Is it a legal document? Is it a letter? Etc, etc. Ditto for pictures, there's a article here that uses some AI techniques to identify pictures by high level concepts, i.e. a picture of a bird or a cat.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Jim Crafton wrote:
I'm sure there are AI projects out there that can scan a text file and give you some clues to it's content
This? http://www.aspenbusinesssystems.com/pages/detpages/xeroxnews909.shtml[^] A friend who is a computational linguist helped write an email categorizer for Inxight, which is now owned by SAP. He's partners in another company that breaks down and categorizes resumes. One issue, though, is that all these assumes correct original information. The irony is that the very people who could use this type of technology would probably have no idea how to even if it existed.
-
Jim Crafton wrote:
I'm sure there are AI projects out there that can scan a text file and give you some clues to it's content
This? http://www.aspenbusinesssystems.com/pages/detpages/xeroxnews909.shtml[^] A friend who is a computational linguist helped write an email categorizer for Inxight, which is now owned by SAP. He's partners in another company that breaks down and categorizes resumes. One issue, though, is that all these assumes correct original information. The irony is that the very people who could use this type of technology would probably have no idea how to even if it existed.
Yeah, that's the idea. Something like this integrated into the OS, so the moment a file is stored, saved, modified, it's scanned and the meta data updated accordingly.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
In 2009 why do we bother with directory names and traditional file names any more? What would be some alternative ways of organizing things? I was wondering about this because my sister-in-law, people I know at work, and even my parents often struggle with remembering what a file is called, giving poorly chosen names to files and then being unable to find them quickly, and so on. I'm sure others here have had to help family members, or maybe even co-workers find "misplaced" files. Yet one of the things I thought was cool about a Palm Pilot was that you never worried about files, or where things were stored. Why can't this be applied to a PC OS? Why not use a combination of things? Something like tagging to add descriptive bits, a UI that has alphabetic index, like a book. Given that NTFS has many of the features in place to do something like this (Alternate Data Streams would be a perfect place to store meta data, tags, etc) how hard would it be to write something that might accomplish this? What other things would be nice to have?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
In 2009 why do we bother with directory names and traditional file names any more? What would be some alternative ways of organizing things? I was wondering about this because my sister-in-law, people I know at work, and even my parents often struggle with remembering what a file is called, giving poorly chosen names to files and then being unable to find them quickly, and so on. I'm sure others here have had to help family members, or maybe even co-workers find "misplaced" files. Yet one of the things I thought was cool about a Palm Pilot was that you never worried about files, or where things were stored. Why can't this be applied to a PC OS? Why not use a combination of things? Something like tagging to add descriptive bits, a UI that has alphabetic index, like a book. Given that NTFS has many of the features in place to do something like this (Alternate Data Streams would be a perfect place to store meta data, tags, etc) how hard would it be to write something that might accomplish this? What other things would be nice to have?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Heck no! At least not on my system. I don't care what crutches other people need. It's bad enough they allow SPACEs in names, and don't get me started on case-sensitivity! :mad:
modified on Wednesday, July 1, 2009 5:20 PM
-
I agree, and I often tell others the same. However the problem persists. And honestly, with all the horsepower that a modern PC has, couldn't we come up with something easier?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Jim Crafton wrote:
And honestly, with all the horsepower that a modern PC has, couldn't we come up with something easier?
Would adding metadata actually be any easier? You need some way of adding the semantic information for the file to the raw data - adding in the human interpretation of that data. File path, metadata - it all takes human effort - can't (with current (software) technologies) be done mechanistically IMO. I would love to be proven wrong, however :-)
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
-
Jim Crafton wrote:
And honestly, with all the horsepower that a modern PC has, couldn't we come up with something easier?
Would adding metadata actually be any easier? You need some way of adding the semantic information for the file to the raw data - adding in the human interpretation of that data. File path, metadata - it all takes human effort - can't (with current (software) technologies) be done mechanistically IMO. I would love to be proven wrong, however :-)
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
Stuart Dootson wrote:
can't (with current (software) technologies) be done mechanistically IMO.
There are some articles on CP and another that Joe mentions (further down) that suggest otherwise. I find it difficult to believe that some basic, though still really useful, analysis can't be done at this point in time.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
-
Stuart Dootson wrote:
can't (with current (software) technologies) be done mechanistically IMO.
There are some articles on CP and another that Joe mentions (further down) that suggest otherwise. I find it difficult to believe that some basic, though still really useful, analysis can't be done at this point in time.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
The items of useful automatic analysis I've seen personally:
- Named facial recognition in iPhoto (I'm sure it's in other photo cataloguers as well - iPhoto's the one I use, though) - works pretty well, but relies on a pretty big corpus of previous, (mostly manually) marked up photos. Still pretty impressive
- Search engines (Google, obviously, and Spotlight).
Other than that, not so much
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
-
The items of useful automatic analysis I've seen personally:
- Named facial recognition in iPhoto (I'm sure it's in other photo cataloguers as well - iPhoto's the one I use, though) - works pretty well, but relies on a pretty big corpus of previous, (mostly manually) marked up photos. Still pretty impressive
- Search engines (Google, obviously, and Spotlight).
Other than that, not so much
Java, Basic, who cares - it's all a bunch of tree-hugging hippy cr*p
Stuart Dootson wrote:
Spotlight
Stuart Dootson wrote:
iPhoto
Seconded! I did not expect the iPhoto facial recognition to work that well. Spotlight of course is a lot better than anything in the windows world.
-
Heck no! At least not on my system. I don't care what crutches other people need. It's bad enough they allow SPACEs in names, and don't get me started on case-sensitivity! :mad:
modified on Wednesday, July 1, 2009 5:20 PM
soyouprefertoreadnameswithoutspaces.
-
In 2009 why do we bother with directory names and traditional file names any more? What would be some alternative ways of organizing things? I was wondering about this because my sister-in-law, people I know at work, and even my parents often struggle with remembering what a file is called, giving poorly chosen names to files and then being unable to find them quickly, and so on. I'm sure others here have had to help family members, or maybe even co-workers find "misplaced" files. Yet one of the things I thought was cool about a Palm Pilot was that you never worried about files, or where things were stored. Why can't this be applied to a PC OS? Why not use a combination of things? Something like tagging to add descriptive bits, a UI that has alphabetic index, like a book. Given that NTFS has many of the features in place to do something like this (Alternate Data Streams would be a perfect place to store meta data, tags, etc) how hard would it be to write something that might accomplish this? What other things would be nice to have?
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Jim Crafton wrote:
giving poorly chosen names to files
I think that's more of a problem than the traditional file/directory structure.
I hope you realise that hamsters are very creative when it comes to revenge. - Elaine
-
Jim Crafton wrote:
how hard would it be to write something that might accomplish this?
Well WinFS started long time back and was eventually abandoned. So it might be quite hard.
I'm glad that I read through this thread before posting. That's pretty much what I was going to say.
Henry Minute Do not read medical books! You could die of a misprint. - Mark Twain Girl: (staring) "Why do you need an icy cucumber?" “I want to report a fraud. The government is lying to us all.”
-
Gary Wheeler wrote:
because you then rely on him to supply the identity directly
If you rely 100% on the user I'd agree, but with the processing power that we have, couldn't the OS examine the file and generate some initial meta data? For example I'm sure there are AI projects out there that can scan a text file and give you some clues to it's content. Is it a legal document? Is it a letter? Etc, etc. Ditto for pictures, there's a article here that uses some AI techniques to identify pictures by high level concepts, i.e. a picture of a bird or a cat.
¡El diablo está en mis pantalones! ¡Mire, mire! SELECT * FROM User WHERE Clue > 0 0 rows returned Save an Orange - Use the VCF! Personal 3D projects Just Say No to Web 2 Point Blow
Jim Crafton wrote:
sure there are AI projects out there that can scan a text file and give you some clues to it's content
Yeah, but they always end up being domain-specific. They need a hint to get started, or some way to constrain the problem space. They also need to do it efficiently enough and accurately enough to be useful, which are probably the real tricks here. There's always simple brute force methods of matching parts of the text to a mass of categorized data (wikipedia, for example), but these are expensive in time and storage.
Software Zen:
delete this;
Fold With Us![^]