There is an article in an old Byte magazine on analizing data to differetiate items in the fewest possible number of queries. This would allow a "product selection wizard" to be generated from product data with keywords. Any idea what issue that was?
To suggest other products related to an inital selection, create a new database called related and when any order for more than one product is placed, check for and create if not found, a records for each product relating to the other products with a "frequency" field that is initially set to... 2? Keep this db size under control by reducing the frequency count on any existing related records for each product ordered where the relation is not validated by the current order and delete the record when the frequency count reaches zero. Allow some flag for records that are always to remain related (i.e. products that have been manually set to appear as recomendations.)
Standardized variable names for shopping cart data
Identifying the computer used to place an order
Price "Fixing" and identity theft in shopping carts.
Displaying a
graphical
price from a price or
part
number. The ability to translate a part number into a displayed price,
removes the need to update the price in each place where the product is
displayed. The "ripple" effect is caused by the method used by the script
to reduce the amount of character image data needed by building the
image from many seperate
GIF
image descriptors
each with a different offset.
See also: