A couple of pre-purchase questions. I don't think these will affect my decision. It is great to find something that can replace the hours I've spent over the last several months :)
1. Is it hard to change the product matching algorithm to go by brand and model?
2. Do you support category hierarchies Electronics --> Televisions --> HDTV --> 36", etc. and if so will a 36" HDTV Television get returned in all three categories ? That is what I am hoping i.e. to have drilldown capability and every additional drilldown would result in fewer products.
Thanks,
Patrick
Hello Patrick,
Taking your questions in reverse order;
The script as it stands does not have support for hierarchical category information. This is intentional i'm afraid, as there just isn't the quality of data out there to support it, leading to unsatisfactory results. However, a popular alternative is to run multiple installations of Price Tapestry in sub-directories, one for each master category; and then you can create a search form at the top level that allows your users to search in any of the top level categories. You can find the code for doing that in the following thread:
http://www.pricetapestry.com/node/205
You might also find it convenient to combine this with the "Drop Record If Not (RegExp)" filter that would allow you to use the same feeds in different installations, whilst only importing specific categories within each one:
http://www.pricetapestry.com/node/1551
With regards to changing the matching algo by brand and model; this is not something that is immediately straight-forward as the entire script is built around comparison by product name (even though brand / category are additional fields in the DB). What you can do however is use the Product Mapping feature in conjunction with carefully registered fields and possibly filters to make sure product names are as similar as possible. For example, if your feeds have both brand and model number as separate fields, you could register model number as the "Product Name", and then prefix %BRAND% using a "Text Before" filter (docs here) to make common product names...
Hope this helps!
Regards,
David.