Forum Moderators: open
Other companies have done one or the other, but never both. Altavista, Infoseek, Excite, and Inktomi have all offered search software before. This is the first time a major company with a history of search has put the duo together.
According to clients, Google is offering the appliance in two models:
one that is capable of indexing 200,00 documents that sells for under $19,500 and a beefier model with a million page plus capacity for $225k.
Evidently, Google has embargoed this story until sometime this week.
If their solution delivers on its promise, this can help keep their free web search free.
Black Box = Can O' Whoop A#$;
Greektomi
I said a while ago that 2002 is going to be an exciting year. :)
Can/Will it be used for web indexing?
I'm not sure how this will effect us?
z6
What I find interesting here, is that the recently introduced indexing of office file formats suddenly starts to make much more sense in this new context (though I still hope they'll eventually put it on a seperate tab on their web site).
Mmmmm... seems like this could give some insight into the algo. Though, it's not clear to me what Pagerank would have to do with a corporate database.
It seems likely to me from the prog contest that google software is written in C/C++. This means that the box probably has only compiled code, hardly worth examining.
Well, pagerank could work just fine. It's just a very small set of page's so you have to tweak some parameters to let PR work (If the algo is smart, i should tweak it along the way, like #referrers needed for PR 8 etc)
I wonder if you could do a mixed search: Search local & internet docs en merge the results.
Yes, but I'm sure that you could learn to tweak the parameters/crawled URLs etc to come up with a pretty good empirical testbed for a limited subset of the web.
For instance if you were in a number of sectors were the Google SERPS returned approximately 150,000 (but no more than say, 500,000) results, and you could learn to set your "black box" up to get at least the top 50 or so pages matching Google, you could test several versions of your new, upgraded site to see what works best.
You could maintain multiple datasets, 1 per keyword/phrase that you wish to test, and offer your best/richest custmers better Google analysis for $500/year.
You only need 40 sign-ups (or 20 for 2 years, whatever) to more or less pay for it, and you'll likely learn something about Google into the bargain. Thats got to be worth it for the real hard-core SEO pros, surely?
Brett, I've been involved with enough big companies to know the right hand usually doesn't know what the left hand is doing. I think it's very unlikely this intranet search box will be running the same software as the Internet search, or at least not for long.
For one thing, I would think an intranet search algorithm would want to replace spam filters with snoop filters. It probably wouldn't make a big hit to trash the CEO's newsletter while caching his private e-mail to his secretary.
I really hope this product gives Google a fat new revenue source. Then maybe they can become more competitive with their Adwords.
Snoop filters sounds like a kind of censorship. Maybe a tool for the managers that returns all suspicious queries. A Big-brother filter. With some creative hacking that Google box can be so much more than just a search tool ;-)
Just my 2 cents now if I just had 2,000,000 more I could find out for myself :)
Greektomi
Is this their core competency? Wouldn't it be better to spend the research dollars on better search results, more understanding how people look for things, etc.?
999,996 cents to go;
Greektomi
Q. Will the Google Search Appliance work on my external web servers?
A. Yes, you can search all of your external Web servers.
Q. Will the Google Search Appliance work on my intranet?
A. Yes, you can search all of your intranet Web servers.
Hope this helps,
GoogleGuy
I've created a few versions of search engines for a limited corporate database. The difference between a corporate database and the internet is there is no spam in a corporate intranet.
Greektomi
I am sure there are safeguards and the like in place to keep this from happening: the algo itself, for instance, is probably crafted in such a way that it might not even be possible to spam it. Further, not being able to see how your page ranks compared to the entire google database would be a show stopper as well, but I would be disappointed to see this cause any problems.
It really does look like Google in a box (a really cool looking box at that). What kind of safegaurds might be in place to keep the kind of abuse I described from happening? GoogleGuy, your thoughts?
a) they have taken precautions
b) soon google.com will have a new algo
Will Google's PageRank™ scoring method work on my intranet content?
Yes. PageRank is effective on intranets, as well as external sites.
I thought PageRank™ is only effective for google.com or hosted site search? If I am indexing 20,000 pages what would be the advantage of knowing how this pages are interconnected internally. Can I download my site's page rank info (if it is available ?) from google.com or am I far off ?
Also does the spell check work for all the supported languages?
Anyone know which 28 languages are supported?
