Google Forum

Double Your Adwords Profits in 7 minutes!


$7 Secrets
People are making thousands of dollars using the techniques and scripts included with this popular report. And it costs only $7! [Read more...]

AdWords Secrets
Free 5-day course that can help you make money using Google AdWords.

Articles for 25 Cents Each
Get 400 exact keyword optimized articles, delivered to your inbox, every month. A new and inexpensive way to build keyword rich web sites that can make money, month after month.

Trade Links with 5,000+ Sites
This is not an automated link system. The Add URL Directory is a directory that lists over 5,000 "add url" pages that have forms that you can use to add your site and exchange links. The directory is divided into categories, helping you find sites in your topic area.

Backlink Analyzer
Automatically analyze the anchor text of all of your backlinks. Analyze your competitors' backlinks to see how you compare in the anchor text department.

SEO Web Site Templates
Web site templates that have been designed specifically for search engine optimization (SEO).

My sponsered child, Hama from Niger, Africa
A portion of the proceeds from this site help sponsor Hama from Niger. Learn more about Child Sponsorship.

AdSense is about search


Author Message
GoogleGuy Says

PostPosted: June 18, 2003 6:07 AM 

Importance: Medium

GoogleGuy explains how AdSense is about search, indexing pages in order to serve relevant ads - the entire page is treated as a search query.
GoogleGuy Says: [Link to quote]
bcc1234, I think I understand your point, but this is completely about search in my mind. Targeting ads based on page content and context is a *really* hard problem, both conceptually and computationally. Instead of doing a search based on <10 words that the user carefully selects, the entire document is treated like a query. You've got to find the salient keywords and features while discarding the parts of the page that don't matter as much. You're searching over ads instead of over web pages, but it's still a hard matching problem--because you have to analyze an entire document.

So I think it's a natural progression of search technology to analyze larger chunks of data like entire web pages. There's some serious technology and semantic analysis that goes into this problem. I don't mind telling you that the engineers working on this problem were using an incredible number of machines to crunch through their models. And we were fortunate to pick up the smart people at Applied Semantics to pool our brain power on tackling challenges like this.

So I think I see your point, but I hope it makes sense that doing queries using entire documents is a natural extension of our search technology.

vitaplease, that's a good suggestion--I'll pass it on.

Okay, I was staying up late tweaking things and playing for a demo tomorrow, and I got pulled into this fun discussion--but I really shouldn't be up past 4 a.m. That's late to work even for a Googler. :) I'll catch up on this thread tomorrow. G'night!

Subscribe to this discussion: Email

Join the conversation:

Remember personal info?

Check to Subscribe to this Comment:
(email field must be filled in)

Subscribe Without Commenting