Danny Talks Google - Day 1, 4:36 PM (EST)

[Garrett]

The session on Search Engine Research just wrapped up, and I spent some time talking with Danny Sullivan (http://www.jupiterevents.com/sew/fall03/dsullivan.html), editor of SearchEngineWatch (http://searchenginewatch.com/).

Garrett: What direction do you see Google's algorithm going in? What are Google trying to provide?

Danny: They're trying to provide more variety in listings, not just commercial but more informational types of websites.

Danny also believes that a lot of people misunderstand the idea of what content is. Website owners typically don't provide content that is truly useful to their market area. "People need to provide stronger, better content," he says. "I want to go to a site and spend ten minutes there, not just go in and see a product page and leave."

What direction are searches moving in? Danny sees searches moving more towards a niche area. For example, let's say you want some news information on, say, Michael Jackson, you could do a search on "Michael Jackson" but you will probably get a bunch of Michael Jackson sites. Danny believes search engines are evolving to the point that they'll be able to tell whether the terms you're searching for are for news information or for a specific website. "That way," Danny says, "if it's something in the news you seek, they'll know to give you news and then have suggestions for sites on the web. At least, that's the general idea."

Look for more comments from Danny in future posts!

[thusmann]

Google has already developed systems for integrating the ODP RDF dumps, so use of such info in conjunction with the algorithms is inevitable (if not already practiced).

I believe that the use of RDF becomes more common, it will become a more important factor in determining position. It should enhance Google's ability to value inbound links by accurately determining the subject and relevancy of the site that is serving the link.

Garrett should ask the Google reps if RDF is, or will be, used by Google to help organize content and results. And if so, what resources would they recommend for learning how to appropriately structure RDF data on websites?

for more info about RDF goto
http://www.w3.org/RDF/

Here's a quote from a Stanford Paper. Hmmm, I wonder if Google is their testbed? ;)

"However, we are heavily exploring the relation between natural language processing techniques and underlying ontologies, and have created a system that is able to continuously learn from what it has previously read and to directly produce data in Knowledge Representation languages from natural language text." ftp://ftp.ksl.stanford.edu/pub/KSL_Reports/KSL-03-07.doc

I think Google is (or will be) using this technology to crawl websites and turn them into RDF format, to promote a smooth transition to the Semantic Web way of doing things, AND to store website data in a manner that permits PageRank/positioning algortithms be applied. Which is how they populated the Froogle database initially.

If this is true, Googl may be using published RDF formulas, in conjunction with their Page Rank formula to determine the position of a website in their results.