This month, Yahoo was granted a patent that describes how link texts may be used to increase the relevancy ranking of a web page. Although the patent was filed in 2002, it provides interesting details about how search engines can use link texts.

Why are link texts important?

All major search engines pay close attention to the text that is used in links pointing to web pages. A web page might be considered more relevant to a search term if the search term not only appears on the web page but is also used in the text of the link that points to the page.

Web pages can even get high rankings for a search term if they don’t contain that search term. It’s enough that many websites link to the page with the search term in the link text. This has been demonstrated with several Google bombs.

How do search engines treat link texts?

While it is clear that link texts are very important for search engines, it’s not clear how much weight search engines assign to a link text when they index a page. Are some link texts more important than others?

The patent indicates that link texts are broken into parts, called tokens:

Once an anchortext phrase is identified, it is converted into a set of tokens. For example, page 306 contains the phrase “best Louis Armstrong site” pointing to page 200. The tokenization produces the following tokens:

* “Trumpet”Best Louis Armstrong site”
* “Louis Armstrong”
* “Louis”
* “Armstrong”
* “Best”
* “Best Louis”
* “Best Armstrong”
* “Best site”

The search engine algorithm calculates a weight for each of these tokens. If the weight exceeds a threshold, the linked web page may be indexed under that token.

Click here for part 2 of this patent analysis.

Popularity: 11% [?]