Next: AltaVista
Previous: A Three Cornered Fight

Algorithms

An algorithm is a procedure or formula for solving a problem. How do I know that? I went to Google, currently the World's favorite search engine and used the define operator

define:Algorithm

Google responded with a set of definitions for the term along with the web pages where it found those definitions. But what really happened when I hit Enter? Google certainly didn't scurry off around the World Wide Web looking for that information while I twiddled my thumbs. In fact I didn't get much time for thumb twiddling as the response came back within a second. Instead Google had already searched many billions of web pages carefully sifting through and evaluating their content. Based on my query it was thus able to determine what pages it thought most relevant.

The way that Google sorts pages for relevance is determined by the algorithms it uses. These examine the text in the page, in the instance above the presence of the keyword: Algorithm would certainly be pretty important. Continuing with this example the algorithm designer may also guess that definitions occur in alphabetic lists with the keyword starting a paragraph followed by some text giving the definition. This could form the basis of the ‘definition’ search algorithm used by a search engine.

Computers are poor compared to humans at understanding the meaning of web pages (semantics) and other documents so most search engines also use clues left by humans. The person who wrote the page and people who link to the page. The page author will use tags that are part of the Hypertext Markup Language (HTML) to guide readers. Important topics will begin with a Heading tag, most end users will see this as a larger font. Other web authors may find the page so useful that they link to the page, possibly with relevant anchor text. The search engine can use this information to rank the page compared to others containing the same keywords.

The exact algorithms used by a search engine are a closely guarded secret and evolve with time in order to improve the results offered to users. Search Engine Optimization experts attempt to reverse engineer a search engine's algorithms. They do this by examining search engine results pages (SERPS), by setting up test pages to observe the results and by exchanging intelligence with other SEO experts on forums or by private contacts. From this information they can infer what factors are significant to a particular search engine. This is done to improve the web sites that they are optimizing (white-hat SEO) on the basis that a site that a search engine prefers will also provide a better user experience. Sometimes they find flaws in the mix and implementation of search engine algorithms and they can use this to unfairly boost the rank of their sites over other, possibly more relevant pages (black-hat SEO). An example would be Google’s use of inbound links to rank a page. When Weblogs or Blogs first appeared SEO experts exploited the most popular to write spurious entries in the comments fields pointing to the sites they wished to boost. They also set up link farms and secondary sites simply to link, and boost the value of their main site. As a result the Google Ph.Ds update their algorithms to put less weight on this information and even to ban sites using strong black-hat techniques.

Algorithms

See Also