Information Technology Reference
In-Depth Information
most “relevant” sites appear first?
With billions of Web pages stored and indexed, search engines of
ten identify thousands of pages that might be relevant to your search
query. Finding the most “relevant” is indeed a challenge, but search
engines want to return to you the most helpful sites so that you will
continue using their service and they will continue to profit from ad
vertisements. In producing their listings, search engines do not want
their selections to be unduly manipulated by commercial sites.
Commercial sites often want to be listed first, where they will have the
greatest visibility and thus the greatest likelihood of being utilized. As
a Webbased company, they may have a wonderful site, but few folks
will visit that site if it appears on page 100 of a search engine's listing!
Sometimes companies are given the opportunity to purchase space on
a search engine, but often they must rely on the relevance of their site
to the user's query. Thus, commercial companies may try to refine
their Web sites so they appear most relevant or important. (We'll dis
cuss how they work to achieve higher visibility later in this section.)
This commercialism leads to a continual tension between orderings by
search engines and manipulations by commercial Web sites. Were a
commercial site to know exactly how a search engine orders its mate
rials, it could try to alter its content to include everything that would
make it listed first. Consequently, a search engine company rarely
makes public many details about just how it determines the order of
its search results. On the other hand, even with details buried within
trade secrets, some common approaches such as the following five are
well documented.
PageRank: PageRank, a registered trademark of Google, deter
mines a page's importance by considering what pages link to it and
the importance of those pages. At a basic level, if many pages refer
to a given page, that page is considered relatively important; if few
pages refer to a page, that page is relatively unimportant. At a more
sophisticated level, a page is considered important if many impor
tant pages link to it; a page is less important if only a few important
pages link to it or, alternatively, a page is less important if many
unimportant pages link to it; and finally, a page is unimportant if
few pages link to it—either important or unimportant. The inter
ested reader will discover the computations involved in this process
