Sphere, the Relevant Blog Search

32 Comments

Blogs are big. There are 19 million blogs, and a few thousand will go online by the time you get done reading this. To paraphrase a popular disco ditty – so many blogs, so little time. Technorati, Feedster and Icerocket do a good job of searching the blogs, the information that is presented to the end-user is like giving people the cow, not the steak. The big issue is finding relevant and intelligent blog posts on a specific topic, that are based on authority. The authority is not an arbitrary decision of a human community, but a “collective effort.” Google tries to do this in its new Reader by restricting “google search” to RSS feeds. But that’s not the answer.

Instead, I recommend you guys take a look at Sphere, a stealth mode start-up that has developed some interesting algorithms to solve this problem. They are in Beta, but I have been kicking around the product from the day the company started in April 2005, and have been solidly impressed by the results they bring up.


More on that in a bit, but here is the skinny on the company. Sphere is the new name of Yodel Search, and it is a three person start-up backed by the likes of Doug MacKenzie and Kevin Compton, Phil Black, Will Hearst, David Mahoney, Vince Vannelli and Mike Winton.

The team has built the product with less than $200,000. The company is headed by Tony Conrad, who is a reformed VC who in past life had funded companies like OddPost when he was a general partner at VSP, a venture fund that has recently imploded for many diverse reasons. The other two co-founders were the duo behind Waypath.

I have digressed enough. The way Sphere works is a combination of many tracks. Lets use an example say of what else, Broadband. The look for blogs that write about broadband, (including those with broadband in the title of the blog) to create a short list. If I am linking to someone who is also a broadband blogger, and vice-versa, Sphere puts a lot of value on that relationship. The fact is most of us broadband bloggers tend to debate with each others.

Think Blog Rank, Instead of Google’s Page Rank. The company has also taken a few steps to out-smart the spammers, and tend to push what seems like spam-blog way down the page. Not censuring but bringing up relevant content first. They have pronoun checker. Too many I’s could mean a personal blog, with less focused information. That has an impact on how the results show up on the page.

The coolest feature they have is matching Blog content with relevant web articles from mainstream media. I believe they are about to launch the beta version of the search tool soon, so try it out and make your own decisions.

32 Comments

my Cancer Community

Hi,

I’m looking for a chart that I thought was published at your site showing the breakdown of the types of blogs e.g., diaries, politics, music, photos, etc.

Any help appreciated.
thanks,
Steve

MikeH

I have had a chance to see Sphere in action during its pre-Beta phase, and have to say I was pretty impressed with their ability to surface relevant posts. I think this is one worth watching; the team has some very credible chops when it comes to building authority and relevance based algorithms.

Brian Shields

“I” am so toast.
“I’ use “I” a lot.
Sphere is going to see all my “I”‘s and rate me low.
“I” Caramba,

kaveh

Did you know that this technology was available back in 1999? The guys behind the Waypath were involved, but not as active employees of the company: Fizzylab was the name of the company and Relevator was the product (there were others like Ad Relevator, Product Relevator, etc.). The company couldn’t close it’s B round of funding and closed its doors … perfect example of baby getting thrown out w. the bathwater. during the crash.

The algorithm has supposedly been enhanced since, but essentially does the same thing, which is much more powerful than keyword search.

Fizzylab’s target customers were publishers on the web. When a user was reading an article, an iframe on the page would show them ‘More like this’ links from the publishers universe of content.

Anyway, if interested, lookup Fizzylab on WayBackMachine: http://web.archive.org/web/*/http://www.fizzylab.com

Perhaps, we were just a bit ahead of the need back then.

Ben

Wow…that algorithm sounds exactly like the one my company uses to data mine relationships for our recommendation engine. Find users who like say Robbie Williams for a short list and then check what artists they like (which gives the uber popular ones as well as others down the curve) then use backlinks to verify those links and create a strong relationship from this.

java2king

there you go Om..now why are you pimpin around a product you are beta testing for the folks..getting ’em publicity in the web2.0 bubble..LOL…kidding…….

Comments are closed.