Desktop Search Engines Compared

nutterButter writes "After Google created a stir with its desktop search engine, other engines gained more awareness in the public eye. Slate did a comparison of them and Google was not their top pick; Copernic was. I tried it - and am quite impressed."

Comments

Linux anyone?

[ewanrg]

Is it too much to hope someone might build a strong tool for doing this that will run on Linux? Having Copernic rated #1 is wonderful for folks still running Windows, and Google is wonderful for folks still running Windows, and...

I assume you get the picture :-)

---

Yeah, I'm like this on my blog too ;-)

Re:Linux anyone?

[QuantumRiff]

No, its got a pretty interface. To be accpeted by the linux crowd, it needs to be "GREP" with a combination of C, perl, shell scrip, and awk. Oh, and better be availble in RPM, tar.gz, and .deb. And it surely better use MySQL as a backend, with apache as the gui (if your going to have one.) We unix geeks like to demonstrate our knowledge by always doing things the hard way!

Re:Linux anyone?

[ken_devon]

Wow. The timing on this article is uncanny. I installed Beagle yesterday, and I'm already addicted to it - it indexes documents, mail and web pages as they're accessed, and updates it search results in real time.

Re:Linux anyone?

[theantix]

To tell you the truth, I'm very glad that these sorts of companies don't yet write software for Linux. A free software solution like Beagle comes without spyware, doesn't send your information to their corporate masters, and doesn't shove ads down your throat or charge you money.

Someday I'm sure that these crapware vendors will be producing their garbage for Linux, and dumb Linux users will be plagued with much the same sort of problems that windows users suffer today. It's almost a golden age now, knowing that the vast majority of Linux software is truly free libre software instead of the ugliness that freeware software will bring.

Re:Linux anyone?

[rusty0101]

Actually, no they don't use a recursive grep on your hard drive.

They use several filters to build an index of words in the various documents they have filters for.

When you ask Google Desktop, Yahoo Desktop, or other search engines to find documents that might be relavent to your search string, they compare the words in your search string with the words in the index they created earlier. From that index, they then provide you with a list of files on your system ranked by whatever algorythm the developers came up with.

If you happen to have a DVD ISO file on your system somewhere, copy it to a different partition to see how long just copying, not searching, that much material takes. It is not a non-trivial amount of time. Especially when you are looking to present a user with a list of matches in under a second.

Indexing is not just running a variation of 'grep' against your files. It is collecting a list of words from each document, identifying those words that are not 'common' (if, and, but, the, or, a, I, etc.) and identifying where in the document those words exist.

That way when you look for 'President Bush' on your hard drive, it can compare the proximity of the words 'president' and 'bush' and give a better match to those documents that contian both words, closer together. That way your disertaion on Teddy Roosivelt hunting in the deapest affrica will be less likely to come up with a match than your discussion of the relaventce of the first Gulf War to political dinners in Japan.

There are a couple tools out there that provide some of these features for Linux. You can use ht://dig to build a web based interface. If you would rather be able to use either a command line search, or a web based search, you might want to look into Glimpse.

Of course, this being Linux, dozens of people have taken a partial stab at doing this. You could probably work out a method from either the Learning Perl, or Learning Python books, as both are quite capable of building and maintaining indexes. The best part is that it would be optimized for your set of files, rather than just being a generic tool that you have to go out and find third party filters to make use of.

Then again, what do I know. If you think running grep against /dev/hda is a good use of your time, more power to you.

-Rusty

Re:Linux anyone?

[dAzED1]

I was poking him since he didn't understand. You don't understand either, but you're closer. As a dba, I'm quite well aware of how indexes work.

If you're organized, then your docs will be on one general area. As such, running an egrep in there for a phrase really doesn't take much time at all. 20 minutes? hardly. A second, maybe 2. Try it some time.

What it allows me to do is make my /own/ algorythm for what I want displayed.

Is this practical, or even easily plausible, in windows? No. Does everyone know regular expressions? No. Am I saying that no one should use these tools? No. I'm just commenting on the poster that said grep couldn't do what these tools do - they were wrong.

locate doesn't search your emails, nor let you know which files containt things, you could recursive grep, but that doesn't find stuff in pdf files, and takes up a ton of cpu.

Locate - doesn't need to search my emails. gmail does that just fine. Egrep tells me what contains whatever I want. Can google's tool find files that have a line that starts with a number, has 2 words, then repeats the number again? No. Simple regex can blow away anything the google tool can do. I can most certainly find stuff in any binary or doc file, without taking up "a ton of cpu."

See? not saying my way is better for everyone else. Just saying someone who says my way doesn't work, is wrong - my way not only works, its more powerful.

Apple's coming out with something like this...

[bennomatic]

It's called Mac OS X Tiger. If you've used iTunes, you know how good and how fast searching can be. It's going to be pretty awesome when it comes out.

Re:Apple's coming out with something like this...

[Shanep]

It's called Mac OS X Tiger.

Actually, it is called Spotlight.

Which will be a part of Tiger, the latest upcoming version of Mac OSX.

history search

[FrenZon]

The biggest use (and what makes it a necessity for me now) I have for a desktop search tool is searching for a webpage I partially remember visiting a few weeks ago, but need more information from. GDS indexes the content of all pages as you visit them, making finding them relatively easy - as far as I could tell (tested over half an hour), Copernic only indexed title and URL, which was of much less use.

A minor point for the geekier here - GDS can also be activated using quicksearch URLs from IE or Firefox, which is handy for those used to getting everything from one field.

the main problem i had with google

[jeff munkyfaces]

is that i can only open the file i search for!

i planned to sort out my music collection - so i searched for an artist - 87 results.

can i select them all and move them to a folder in one go? no.

for this kind of thing it's useless - i wonder if i can with copernic..

Some GNOME folks look to be working on it.

[Chuck Chunder]

Beagle is a search tool that ransacks your personal information space to find whatever you're looking for. Beagle can search in many different domains.

The latest edition of the Beagle newsletter has just been released.

Re:Some GNOME folks look to be working on it.

[artemis67]

Whoa, whoa, whoa. If you're talking about having your personal information ransacked, then Windows wins again!

Re:Why would anyone trust this?

[Old Man Kensey]

Assuming you're the kind of user with privileges to install this on your computer to begin with, every application you run already has those permissions. Any program you run has the same permissions you have when you run it, unless there are admin policies to the contrary in place. So, these apps don't inherently represent any more of a security risk than the ordinary search built into your OS.

So, do you trust your OS vendor? If so, why, exactly? For that matter, do you really trust your antivirus vendor?

Re:Why is desktop search so hot?

[mOoZik]

Actually, it CAN search inside of files, contrary to your post. The results can then be arranged by size, type, folder, date, etc. Isn't that enough?

Newsflash

[E IS mC(Square)]

slate is already sold to washington post. . and here

Desktop search unnecessary

[Dogtanian]

All desktop searches are redundant; well, under Windows at any rate.

Simply use Google, which will have visited the web server on your compromised Windows PC- the same web server that is sharing everything on your hard drive with the rest of the world.

I bet those Linux weenies are jealous now.

FYI, Copernic contains adware.

[Shanep]

Copernic's Privacy Policy reveals that, "Copernic Technologies, Inc. works with third parties that transmit advertisements to the Copernic Agent and Copernic Desktop Search product families and Copernic Meta."

Take your pick

[useosx]

Please, the Mac shareware developers practically invented this genre:

Launchbar (the first)

Quicksilver The current favorite, and free.

Butler About the same as Quicksilver, more features but not as slick.