Slashdot Mirror
Microsoft's Search Engine Plans
prostoalex writes "Andy Beal from SearchEngineGuide.com interviews Robert Scoble from Microsoft. Scoble tells the audience what current search technologies Microsoft is working on as part of its Longhorn/WinFS development as well as in the field of Internet. Scoble also discusses current problems with local drive and Internet searching, such as absence of metadata for a lot of files out there: "When I take pictures off of my Nikon, they have some metadata (for instance, inside the file is the date it was taken, along with the exposure information) but that metadata isn't useful for most human searches. For instance, how about if I wanted to search for "my wedding photos?" Neither X1, nor Windows XP's built in search would find your wedding photos. Why? Because they have useless names like DSC0001.jpg and there's no metadata that says they are wedding photos.""
Make it "Micosoft Search powered by GOOGLE". Then "maybe" it might function well. Also metadata needs to be created by the user, I aint gonna be entereing data on a keypad on my camera for every photo.
"It's so convenient to have a system where everyone is a criminal" - A. Hitler
For instance, how about if I wanted to search for "my wedding photos?" Neither X1, nor Windows XP's built in search would find your wedding photos. Why? Because they have useless names like DSC0001.jpg and there's no metadata that says they are wedding photos."
That's why you can change filenames and organize things into directories.
I think thats what "organization" is for. You place files like "DSC0001.jpg" in things called "folders", and then name the folder "Wedding" or something.
I dunno.
Bowie J. Poag
What happened to thumbnails?
So instead of offering their official toolbar for IE only (the one for Mozilla is unofficial), start to slowly phase out the Google Toolbar and replace it with the Google Browser which would basically be a Google branded Mozilla Firebird. With all the features that make Firebird great like Tabbed Browsing, with the addition of the Google Toolbar features such as PageRank, etc. All on a cross platform basis.
If people get used to downloading better browsers now, then they won't even notice when the next release of IE starts to reject the Google Toolbar.
Let them know what you think
If users didn't suck so much, then descriptive dir names would easily solve the problem of trying to locate a wedding photograph on a hard drive.
So what, the image file is named "DSC0001.JPG" -- who cares. Put it in a folder named "my images" and there's no wonder you can't find it!! Put it in a folder named "wedding photos", and then you've got something there!
The best way to describe it to the average joe (non)user is that directories/folders are analogous to folders in a filing cabinet. Would you file telephone bills, for example, under "mortgage" or "telephone"?
Thanks Microsoft for "my photos", and "my documents", and the like. We appreciate it!
Skiers and Riders -- http://www.snowjournal.com
I hope the industry sees the opportunities that Longhorn's WinFS opens up. We can either work together and share data with each other, or we can be afraid and keep data to ourselves.
Share data? with whom? how can you share data that is in either proprietary format or "patented XML" ???
It is following the OpenStandard that will help in "working together and sharing data".
Consensus is good, but informed dictatorship is better
Why? Because they have useless names like DSC0001.jpg and there's no metadata that says they are wedding photos.""
d aylight...horizon?), you've got more issues than just locating a particular photo.
You mean to say you don't know the date you got married? You're in trouble.... iPhoto on OS X at least breaks them out into folders according to either last imported and/or month/year etc.. You're responsible to breaking them down further, in which case you don't search the entire drive later, you simply open iPhoto and take a short trip to your wedding folder, just like having a folder in a drawer in a cabinet in your home.
It's not really that hard, now is it? if you're dropping any files onto your drive randomly, the issue is with your basic housekeeping, not that a top level search tool seems blind to your target.
You're talking about EXIF, and the list of data there is long. Why you took the picture isn't part of it, and if you want the camera to interpret which part of the subject matter is root (noses..faces...age...sex...background..,trees...
Would you want to trust your private data, gathered from govenment departments, purchases and financial transactions etc, being accessed by such a system run by any old govenmental or business agency?
How about your private correspondence on friends and acquaintances home computers.
Microsoft culled the URL name:password@ functionality from Internet Explorer because it claimed it could not create a secure enough fix, yet in the same month, it yet again proposes a privacy nightmare such as this? Madness.
WinFS sounds promising, but unless Microsoft makes the WinFS specs open and free, it'll be yet another lock-in technology, which would be very dissapointing.
Adding metadata to all your files would require a lot of time and effort, and if it's a closed technology, it'd be yet another reason people wouldn't want to even attempt switching to another OS. I can almost hear it now...
"This other OS looks cool, but I've spent so much time adding metadata to all my files, and I can't export that metadata to this other OS because the format is proprietary and patented... I'd better stick with Windows, switching OS's would be too hard..."
Sorry, someone had to state the blatantly obvious. As usual, all promising technologies coming out of Microsoft are poisoned. And most people don't even realize it. Not even intelligent people. Most .NET developers don't even realize that .NET's so-called "standardization" via ECMA doesn't really make it an open standard (lots of the "standardized" .NET technology is encumbered by patents).
-Teckla
For instance, how about if I wanted to search for "my wedding photos?" Neither X1, nor Windows XP's built in search would find your wedding photos. Why? Because they have useless names like DSC0001.jpg and there's no metadata that says they are wedding photos.
:)
You know the date of your wedding right? If not, don't let your wife find out. You can search for jpegs taken on a certain date.
As you previously said...
When I take pictures off of my Nikon, they have some metadata (for instance, inside the file is the date it was taken, along with the exposure information)
It isn't exactly rocket surgery.
LK
"Hi. This is my friend, Jack Shit, and you don't know him." - Lord Kano
But that way, things can only be in one folder. The whole point of the WinFS stuff is that you're not forced to store stuff in a rigid hierarchy.
What if you wanted to find your favourite wedding pictures? Ones with your Mum & Dad in them? Ones with your wife in, but not necessarily from the wedding?
I'm a game developer. We have a lot of "materials" which describe how to shade surfaces. Our editor allows you to put the materials anywhere under a certain directory. That's nice, as it avoids having thousands of files in the same directory, but in practice it's a bit of a nightmare.
They're currently grouped in a number of different ways. All of the artists have their own folder. Then there are folders for each different level of the game. Also there are "grass", "metal", "wood" etc., and they're also sometimes grouped by slipperyness. It's a complete mess, as to find a material you have to check all the categories it might fit into.
Allowing them to be dynamically grouped based on metadata would be a dream. I would implement it myself if I had time.
MOD PARENT UP. A little common sense goes a long way. Entering all that metadata that would make keyword searches viable would certainly help, but people already have the ability to do that - it's not worth the time or effort. Cataloguing by directory and maybe filename is all I ever do and I'm pretty sure I'm not alone.
-N
I've nothing to say here...
"Why? Because they have useless names like DSC0001.jpg and there's no metadata that says they are wedding photos."
Metadata is a stupid concept. It puts the cart before the horse. Files should not have to 'know' about themselves, they are not objects.
You have to treat files as just files, their names are nothing more than identifiers, their contents are nothing more than contents.
By all means its possible to build a great search capability into a filesystem, but you need to build the 'meta' data _outside_ the file.
A system built on file metadata is doomed to be incompatible with anything but the latest datatypes designed for it.
Used to be if you wanted to find a file real fast under windows, you'd hit WINDOWS+F to pullup the find window, enter in your search query, and Go.
Now if you're in front of an XP machine and want to find say...all the pictures on the system you can't just enter in "*.JPG" anymore. You have to read what some animated dog is asking you, click on one of the options before you get to the search query window, then enter in the query. doesn't sound like much of a hassle, but it IS an extra step.
$cat
If your file system supported true symbolic links, your problem could be largely mitigated by using them.
"Give away the stone, let the oceans take and transmutate this cold and faded anchor." - Maynard James Keenan
"Neither X1, nor Windows XP's built in search would find your wedding photos. Why? Because they have useless names like DSC0001.jpg and there's no metadata that says they are wedding photos."
Metadata will NEVER improve searching in this way unless the things that generate the content FORCE you to put it in before they can snap pictures, etc...
Even if people were forced to put metadata into all their files there is a big chance that typos and other errors in entering the info would occur. This will make the metadata totally useless in a search!
It's saved my bacon more than once. As we move away from text, we become completely dependent on metadata to find things. Standards for metadata need to be settled soon, or Moore's law means our computers will become less and less useful.
--Mike--
"As we move away from text, we become completely dependent on metadata..."
Exactly what do you think metadata is? This system would require more text than current. At present, you can rename the files and put them in folders, which works quite well if you have any organizational ability. Metadata would require dozens of unrelated pieces of info be input, and the a more complex retrieval (search) process would be required. While metadata standards are important, it's only advanced users who will be using them. How many "typical" users do you know that are going to search for a photo by the F-value?
And for the record, I've never used the "containing text" search, because I name files in unambiguous ways.
G
If Dylan had a more common name, like, uhm...Mike... the value would go down. What would you do to include the names of the other 4 people in the photo? How do you link it to Dylan's other photos, etc?
--Mike--
Who's really dumping all their pics into My Pictures without making subfolders? Subfolders and thumbnail view in XP already make searching for photos from a command line useless.
Support the First Amendment. Read at -1
This is more useful than it would seem. I've read a bunch of posts that talk about keeping the wedding pictures in a folder called "Wedding", and that's the extent of the organization.
Except it doesn't work that way. If I dig around a little, I see that I have the same images in several places: in the folder called "Vacation", another folder called "work" where I did some touchups, another folder called "staging" where I laid things out before putting them on the server, and again on the server, where my family can view them on the web.
If I follow the suggestion of putting them all into a single folder, then I've created a logistical headache. The _only_ thing I've gained is the ability to find all the files at once. Using metadata, I would no longer have that restriction - I could put files where they made the most sense, and still find all the files at once.
ACDsee, a well-known and, at one time, free, image viewing and organising app, supports metadata. It puts it in a "descript.ion" text file in each directory. This is an ancient DOS standard. It's still supported by a few Windows apps, notably the Far manager (a shareware clone of Norton Commander for Win) and ReGet, a downloader; both Russian.
In fact I find the "descript.ion" metadata so useful I stick with apps that use it. At my last job, a web news site, I organised out image library using ACDsee and this metadata to add notes. ACDsee also has a nice batch rename.
No need to invent a whole bloody new file system to find your wedding photos.
Imagine a world where we're all broadcasting identity. Say we've got RFID-enabled nametags at the wedding. Now picture a camera that has a (preferably directional) RFID reader. Suddenly all your photos have the names of all the subjects automagically added as metadata. Scary in some contexts, useful in others (like most technologies, I suppose).
Cripes. You'd had thought that a company as big as Microsoft would have considered a better way, but no.
.. you got it .. drop it on the Uncle Jim topic icon.
-
from the article, Microsoft's Robert Scoble:
But, WinFS goes further than X1 and other file search tools do today. It lets you (and developers of apps you'll use) add metadata to your files. So, even if you don't change the name of your files, you might click on one of the faces in a picture application and get prompted to type a name and occasion. So, you would click on your cousin Joe's face, type in "Joe Smith" and "Wedding."
So Microsoft, who have sold many more graphical interfaces that anyone else on the planet, require you to "type in" Joe Smith for each and every photo of Joe you have !
Oh, sure, there'll be a dropdown list, but it'll surely list every last irrelevant person and topic you ever defined in WinFS.
Instead consider the following scenario: -
You've uploaded your latest batch of photos from your camera to your PC and have them in thumnails view in a file manager of your choice. -
Now you want to add your metadata, so you open up your "Meta topics" folder and select a number of graphical icons representing the subject matter of your photos, e.g. "Wedding", "Uncle Jim", "Mary-Jane" and some others. You then drag'n'drop these into a "Scratch" folder and close the "Meta topics" folder. So you now have the freshly-uploaded photos, and the relevant meta topics. -
Now select all the photos in the folder - they're all wedding photos, so drag'n'drop 'em onto the Wedding topic icon. -
Now select the photo of Uncle Jim staggering across the reception with a pint of special, and
Now the picture of Mary-Jane in her wedding hat - yeah, that's it baby - drop it on the pretty icon.. -
Now you can access all the Wedding photos by clicking on the wedding icon all the pictues of Uncle Jim by clicking the uncle jim icon and so on. -
There's even an interface to combine filters, e.g. Wedding AND (Uncle Jim OR Mary-Jane), simply by dragging and dropping the icons onto AND and OR icons in a cumulative fashion.
Now you can do all of this (bar the interface combine filters interface) TODAY, albeit in a fairly crude way, with a file system that supports symlinks (such as ext3), and a graphical file manager (say, Rox-FILER..). And here is my claim to prior art in respect of this "graphical metadata manipulation" concept. Of course, I had to hold down Shift+Ctrl to make it do the symlinks when I dropped the photos on the relevent icon, which a proper interface wouldn't require. Also, a posix filesystem is not as elegant as say, a relational database for the purposes of storing the metadata. But hey, not bad for 5 minutes work. How long have Microsoft been working on this exactly ?
$ strings FTP.EXE | grep Copyright
@(#) Copyright (c) 1983 The Regents of the University of California.
Folders are annoying, though, because (on windows) files can only be in one folder at a time. Unix has links, but even then -crucially- you can't introspect on a file to find out what folders it is in. What is needed is a relational filesystem, where the "directory" structure is _equivalent_ to metadata. I can look for a file in directory mp3, or I can look AT a file and see that it is in directories mp3, kittymonkey.
I have given up even trying to use MS search on my hard drive. First of all it frequently desides that I have nothing on my machine. In other words whenever I search for something it instantly decides that it's not there. The only cure is to turn off indexing and delete the database files. Then it takes an half an hour to search for one filename.
I have taken to saving things on a samba share just so I can use locate and grep.
War is necrophilia.
Actually no, the trick is incorporating it into the filesystem LOCKS YOU INTO USING THAT FILESYSTEM EXCLUSIVELY.
Your metadata would no longer relate to an image, it relates to a particular file on a particular filesystem. If you copy that file with anything other than Microsoft Magic Filesystem Aware Software (like, I dunno, a 3rd party FTP client) and your precious metadata goes up in smoke faster than "medicinal" marijuana.
Visit CryptoGnome in his home.
what will this meta data do with older windows versions or other operating systems?
will it lock them out from viewing longhorn only data?
you better bet the farm.
microsoft isnt only locking in the bios, they're locking everything in.
it's like if you read the libjpeg readme, jpeg isnt a concrete format it can be icompatible with itself.
and microsoft will possibly create data on each file that their software will produce and any other programs on the system, or older windows versions or any other system will be locked out from viewing those file besides let's say....Microsoft photoready. (wouldnt be surprised if they came up with that cliched name)
personally, this winfs search crap is another front for lockIN, and really, I hope people react the way microsoft expects them not to do, flock to an alternate OS
not to mention once programs like photoshop need to use this technology to run on windows, and it would hurt their macos support, not to mention they'd have to pay royalties for it, I think Adobe would start shifting away, and many gaming companies would prolly start flocking as well, because they dont want to put up with this shit, I think atm, the way XP crashes (where if you need to poweroff the system, you cant because XP controls the power on and off of the system) is about how far people are gonna take it.
Teacher next to my third period class has had it with his current XP set up because it just keeps getting viruses that cripple the system, and he spendsmore money and time repairing it than actually being able to use it.
and microsoft wants to lock us into this sort of bullshit? riight, once I told him about linux, he started sound a bit interested in it.
You show people something they havent seen before, plus it being stable, and showing general advantages, they will use it. most people are trained to think you have to pay for quality.
hell, this woman was selling a bunch of computer parts for 5 bucks yesterday (got me a nice monitor, hehe) and she said that these old games she had for her children no longer worked on XP, so she just got new ones.
reason people havent flocked over to linux is because it's completely new to them and their neighbor doesnt use it, but your typical grandmother or parent or average joe will get it as long as it works garaunteed. to your more geekcentric user, it takes more convincing.
but to get back onto my original point, microsoft is gambling with this metadata crap.
it's either going to make or break them.
not to mention if another manufacturer offers support for older windows systems and linux and alternatives, epople would gladly flock to that instead of this new system which doesnt work well with their older stuff. though then again, they might just flock to it because it's the latest thing and they're willing to change. this is something I think all desktop linux companies need to keep in mind.
These standards are already available, are already being used in cameras and imaging software,
and are documented well enough that support can be implemented into your open-source imagebrowser or other app.
Unfortunately, they do not all support the kinds of information you may want to store. EXIF may miss a feature, JPEG2000 another one. The smallest common denominator is probably not desirable.
Besides, certain formats do not support metadata that well, or at all. But you may be forced to use those formats anyway. So data has to be stored outside of the file.
The most obvious shortcoming - it's not feasible to search all files for a certain query. The metadata has to be cached in order for the query to be answered fast. That caching can (and should) be done in some sort of database. If it's on the file system level, all applications can profit from it.
The idea is fine (although not that new, really). It's up to MS now to deliver a good implementation for their OS.