Flutterby™! : Spam to gold?

Next unread comment / Catchup all unread comments User Account Info | Logout | XML/Pilot/etc versions | Long version (with comments) | Weblog archives | Site Map | | Browse Topics

Spam to gold?

2009-01-30 14:17:52.209554+00 by Dan Lyke 5 comments

Thought o' the moment: I've been trying to figure out ways to get better search results than Google. I wonder if some of these word structure scorers, like are used in spam filters, would be a good way to look for web pages written by people who write in a style that I believe indicates knowledge?

[ related topics: Interactive Drama Spam Monty Python ]

comments in ascending chronological order (reverse):

#Comment Re: Domain Concordances made: 2009-01-30 14:39:16.348835+00 by: jeff

Dan--are there concordances of specific knowledge domains you're interested in?

#Comment Re: made: 2009-01-30 15:25:14.749411+00 by: ebradway

I believe the machine learning techniques used by Google have been shown to provide better results than any other technique used. Personally, I think some of Google's methods might ultimately be approaching an an asymptote in machine learning and general knowledge representation.

The only way to make search better would be to further filter Google's output based on who looks at what. If I were to enter a search query similar to one of yours, I'd probably want to have the links you clicked on sorted to the top as opposed to most other people.

#Comment Re: made: 2009-01-30 16:17:57.012737+00 by: Dan Lyke

Jeff, I'm finding that when I'm searching for product information, I'm far too often ending up with the bizrate.com style spammers, and having to dig way way down for reviews and user experience stuff.

Eric, same thing: it's a constant struggle between the spammers and Google, and I think the spammers (loosely defined) are winning right now. This may indeed be that the mass of Google's customers are interested in finding where to buy rather than what to buy, but I think there's room for search which is heavily individualized, just as the better spam filters are becoming heavily individualized.

#Comment Re: made: 2009-01-31 10:26:02.613063+00 by: DaveP

Seems to me like that's a very interesting question that might land a guy a job at google if he was looking. Which also seems like it might be the best way to get an answer.

But I'd sure like a way to search for a product and exclude bizrate-type-sites, too.

#Comment Re: Wiki made: 2009-01-31 13:20:21.122471+00 by: jeff [edit history]

Although it's not related to product research, I've been doing many of my focused knowledge Google searches this way, always including wiki in the search pattern.

I personally find persuing KNOWLEDGE via Wiki to be very efficient, due to all of the related wiki-cross-linking and external linking. For example, I never knew that Nikola Tesla and Mark Twain were good friends, and that Twain was acquainted with Henry Rogers. I suppose I'll have to see if the six degrees of separation includes any of my ancestors in heretofore unknown relationships. Unfortunately, Facebook didn't exist in the 1800's...

Here is a product search using wiki in the search pattern.