ABOUT AMAZON.COM
Copyright ©2010 by T. Pavlidis
When Two Books are Related
People who shop at amazon.com are familiar with suggestions such as "You might like these books …" (when you first visit the site) or "people who bought this book, also bought …" (when completing an order). This is all done by a program that keeps track of what books people buy and links books when bought by the same person as illustrated in Figure 1.
 |
Figure 1: Two different customers bought both the first and the last book, so these two books are considered related. (In reality things a bit more complicated because books are marked as related depending on the percentage of common sales rather than depedning on absolute number of sales.) Of course the computer does not draw colored lines between customers and books but it does the equivalent by keeping arrays of data. |
It is customer actions rather than any content analysis that determines, in the eyes of the seller, whether two books are on related subject or not. The following story confirms that conclusion and also shows how the method may fail. I am a member of a book club who decided on successive months to read two books that were unrelated to each other and not particular popular. As a results the sales to the book club members were a significant part of the overall sales and that led amazon.com to recommend each book as related to the other.
There are other sources of error such as buying items for gifts. For example, If the gift recipient is a child an association between toys and, say, history books may be created.
Data Mining
The purchase record of a customer can be used for other purposes besides determning affinity amongst books. My credit card company can find out a lot about me by analyzing my shopping history. Such a process is usually refered to as Data Mining. This information can be used to detect unauthorized uses of a credit card. A large value item is bought that is quite different than anything else that was bought before will raise a flag.
First Posted: May 11, 2010 — Latest Update: May 11, 2010