Wed Jan 23 10:03:10 PST 2019

203
The core of the Teoma search technology is based upon the idea that society and
the Internet consist of tiny communities that self-organize themselves into hubs
and authorities.
Hubs and Authorities
An authority is a site that is linked to by many sites and pages covering that topic.
A hub links to many relevant topical sites. It is said that a good authority has links
from many good hubs, and good hubs link to many good authorities.
If you read search engine papers and information mining topics such as latent
semantic indexing and multi-dimensional scaling, you can learn how some of those
technologies are similar to what Teoma does. LSI works at understanding word
relationships, but Teoma looks at understanding link relationships between pages
within communities.
Search engines create a reverse index of all the terms in their index. For example,
maybe 10,000,000 pages have the word cheese in them. After a user searches,
Teoma will look at the local term space to find similar terms to cheese and the local
communities that surround those topics. Teoma takes a snapshot of the area and
bases most of its rankings off of local interconnectivity of that subset of search
results.
Below is an oversimplified image. Notice how pages that link to the same pages
may be assumed to be related. Also, pages that are linked to from a common page
may also be deemed as related.
In the image, ?E? may or may not be an authority page, depending on what type of
link it is. If a page has a link from only one hub, then it may not be topically
related. For example, many hosts or web designers add ?designed by? or ?hosted
by? links on some client websites.