First, for those who may not be familiar with Apostolos Gerasoulis and the Teoma search technology, ExpertRank (which now powers Ask.com), here's a brief synopsis:
- The web is made up of communities of experts on given topics (science, math, literature, underwater basket weaving, etc.)
- The links within a subject-specific community can be identified and given higher weight than links from off-topic websites or pages.
- Using subject-specific communities of experts as the basis for link weighting will produce better results than a global link metric that doesn't differentiate based on topic.
For a long time, I've felt that the Teoma technology was a strong alternative to Google, Yahoo!, & MSN's ideas of global link measurement (though those engines may very well also use some degree of community measurement). Lately, however, I've been thinking that the entire hypothesis behind Teoma - that topical communities know better - might be false.
The topical community system would have a tough time with many of the emerging (and some of the classic) patterns of recognition and discovery on the web, including:
- Dogmatic Communities - in which sites around a topic may be insular and incestual, prohibiting new content or new ideas from making a significant appearance.
- The Blogosphere - where subject-specific communities are greatly outnumbered by random-topic sites, and even sites inside a subject area often point to resources outside their expertise.
- Populist Intelligence- the idea that the popular vote, rather than the votes of the few in power, will produce better results.
- Multi-Topic Sites - Wikipedia, Digg, Reddit, Newsvine, Squidoo, and hundreds of other emerging properties don't have a topic expertise per se.
- Temporal Trends - Many of the "expert" sites that the Teoma algorithm was built around followed the classic model of static, resource hubs. Today, a great number of sites evolve and update content and links on a daily basis, pointing to the latest and greatest. The resource sites, meanwhile, are often stale and conservative, failing to pick up on trends or new information.
What's your opinion? Is the Teoma concept still viable? Was it built for an earlier model of the web and is now obsolete, or can it evolve and embrace social networking, user-generated content portals, the blogosphere, and less well-defined web communities?
David: It's not just universities and just not books. Most libraries in the US, Canada, and elsewhere offer FREE full text articles from home or office 24x7x365. All you need is a library card. Actually, in some states you don't even need that, you might be able to access via your IP. In other states your drives license works.
2 more points. In most areas you can get library cards from several libraries. That often can be useful since all libraries offer differerent databases for free.
To give you just one example, here's what the San Francisco public library offers.
https://www.sfpl.org/sfplonline/dbcategories.h...
The first databases listed contains about 60,000 articles. You also see the full text of the NY Times back to 1851. These are the same articles you would have to pay 4 or 5 dollars elsewhere. Thousands of new tech books (full text via Safari, and much more. It's all free. A bit more in this article.
https://www.betanews.com/article/Finding_Answe...
gdp - That's Amazing! You can see the full text of the NY Times back to 1851? Really? Wow!!!
This post reminds me of an Aaron Wall post. (Just a meaningless observation from a search purist)
You mean i'm not the only underwater basket weaver out there? Bring it on!
Groups of experts have been a problem for many years. One such group of experts laughed at Darwin when he proposed his evolution theory. Other groups of experts have conveniently ignored contrary evidence when forwarding their own agenda (see global warming). It happens all the time. So why should any group of experts be any different when it comes to the internet.
An expert on nutrition empoloyed by a mutinational is going to toe the party line. An expert on security employed by a software company is going to use the corporate message. And so it goes on. Experts have bills to pay like everybody else, so they are not going to bite the hand that feeds them - they will be expert in following their employers policies.
I could go on but I'm in danger of ranting and that's not very impartial!
I don't really understand how they measure expertise.
I agree that the best place to look for expert content is through the massive .edu repository on the web. I frequently jump over to universities and surprisingly, many of their libraries have detailed expert sources available for free. Sometimes it's just worth it to look up the books on Amazon, buy them used for a couple bucks than to peruse through mediocre "expert" material.
Hi Rand. On the technical issues, I'll let the technical team answer if they decide to.
Regarding expert sites.
There are NUMEROUS expert sites out there where subject experts (REAL PEOPLE), librarians and others are making judgement on the quality of what goes in the database. They have been out there for years but no one pays attention. In many cases they also offer full annotations, subject access, and excellent maintainance.
Here's just the tip of the iceberg.
Global Business https://globaledge.msu.edu/
LII.org https://www.lii.org
IPL https://www.ipl.org
INTUTE https://www.intute.ac.uk
Entrpreneurship https://research.kauffman.org/cwp/appmanager/r...
Research Guides https://lib.nmsu.edu/subject/bord/laguia/ https://www.loc.gov/rr/askalib/virtualref.html... https://www.loc.gov/rr/international/portals.h...
Why now with this new Google tool. Have we been looking at what people have been creating with Rollyo for the past year. We have seen few 2.0 companies that have received the attention Rollyo has including your own 2.0 awards. My guess with the new Google service to be on the lookout over the long term for lots of spam. You'll see compilations with few good sites that gradually work into the spam mode. Just a guess. Like all things Google, only time will tell.
Rand,
I am not too familiar with Teama's concept, but the way I understand I think it is still viable. It just needs to be updated taking the new trends in social media into consideration. However, I think Google and the other search engines will be doing this pretty soon and I don't see Ask.com really making that much of an impact. Search users care about relevence and Google is already doing a pretty good job at that.
With their new custom search engine Google could determine which sites are authority sites by what users are adding to their own custom search engines. I don't know if Google will actually do this but I think it's a possibility.
CS - Great call on Google's use of the custom search engine to find what real "people" think are authoritative sources. Now all we have to do is write a bot that adds SEOmoz to hundreds of personalized setups and... oh wait.
Seriously, though, I do think that whenever Google puts out a product, there's always a data harvester somewhere in the 'plex thinking about what to do with the information.
Rand,
I think that the social networking craze will help fuel and adjust the expert community model. Maybe these expert communities could also use the customizable search engines that now available to create an engine tailored to their speciality. If they make it public, normal folk might come to use them. I think that an engine focused on investing and personal finance has potential to catch on...
There is a great book called "The Wisdom of Crowds" by James Surowiecki.
It shows many examples of how collective wisdom is always stronger than the wisdom of a few...even if those few are "experts."
Makes you wonder why the decisions of crowds in elections aren't always wise :)
Perhaps that's more to do with those crowds basing their decisions on the wrong info ;)
It is the fact that our system unfairly favors incumbents and those with large election chests full of money is what screws elections up...
As you all probably know, there is a version of Wikipedia now out, or out soon (I don't visit Wikipedia that often), which has experts moderating the changes to the articles. In this way, new and different ideas are allowed and encouraged, links are required to be relevant to the article, populist intelligence is controlled by the expert moderator, articles are constantly updated, and Wikipedia remains multi-topic while having indepth articles on single topics. So if there were just 10 more Wikipedias, or if Wikipedia become universal and had hundreds of thousands of experts and hundreds of millions of contributors, Teoma could make ExpertRank based on Wikipedia alone and give more weight to links in relevant articles for related keywords.
On the same line of thought, I was thinking today why Google should by StumbleUpon. StumbleUpon is like a community of search engine spiders. Sure, people won't judge a website the same way a SE spider does, but they give it a thumbs up or down, and they leave reviews, and the sites that have more thumbs up are shown in stumbles more often. If Google or any company bought StumbleUpon, the intelligence of all of those human editors could be applied to accurately weighting links on the basis of the tags people attach to the website, the reviews, the thumbs up vs. down, etc.
I guess it all depends on how they decide if a site should be in a topical community. How it defines a new topic, and a new topical community among that topic, would be the most important thing to know.
It could be very viable, if implemented right. But I think the best search engine will be the one that picks up the idea of topical communities, combines it with all the other ways of ranking out there, and comes with one hybrid solution that takes it all into account. I don't think the search for "the final ranking solution" is over, it has just started...
And seriously Rand, even when you don't know if this kind of ranking works anymore, you surely know that everything knows better than Digg? :)