Information Technology Reference
In-Depth Information
to be an elegant application of a mathematical subject called linear
algebra (involving eigenvalues and eigenvectors).
Review of Title or Keyword Information: Every Web document
can contain a title (often displayed at the top border of your
browser window) and a list of keywords. Search engines can utilize
this information to determine the relevance of a page to your query.
For example, in your search for “virtual memory,” the body of a
page may contain both words, but the title or keywords may indi
cate the main topic involves Hollywood movies. Because your terms
do not appear in important locations, this page may be considered
less important than one giving your terms a higher profile.
Combinations of Words in Your Search: In your search for “vir
tual memory,” the search engine might locate many pages that con
tain both “virtual” and “memory” somewhere in the page. Greater
priority might be given to those pages where these two words are
next to each other, just as you typed, rather than appearing sepa
rately in the document. (If this approach is used in ranking search
results, a search for “virtual memory” may not produce the same re
sults as a search for “memory virtual.”)
Click Popularity: Click popularity measures how many people
click on various links. When a search engine displays pages, it
counts how many times users select which pages. Those pages that
users select are given an increased ranking.
Stickiness: Stickiness measures how long users stay on a se
lected page. If a user selects a page referenced by a search engine,
the user might scan the page and immediately return for another se
lection, or the user might spend some time reading that selected
page. A search engine might work to record this amount of time,
and pages are given high ratings if users stay on them for long peri
ods of time.
Search engines probably use several of these techniques in some
combination. Some materials from Yahoo! suggest that it gathers
data on keyword and title information as part of its ordering
process. Some thirdparty studies suggest that Google includes both
PageRank and keyword/title information. Click popularity and
stickiness are reported to be used by Lycos, Hotbot, and MSN
Search. Undoubtedly, search engines use other factors as well, but
these give a general idea of the types of strategies available to search
engine companies.
