That does sound like a challenge. I thought of two possible ingredients:
1. Require unique descriptions. This could be computationally intensive, especially at the beginning when you need to deal with already published content.
2. Have pages of standardized product descriptions that can be linked to individual suppliers (and be linked from them, too) if any supplier doesn't feel they can do the work of generating unique descriptions.
Not if you link to a single description from the product pages ... Rather than putting the description on each page itself, you could put the description on a single page and link to it from the product page ... This could also be a good time to use an iFrame.
You could also possibly reverse what you are currently doing and rather than putting the description on all the product pages, put all the products on a single description page.
Normally product description is quite long. So I randomly select 2 lines from every product (each product throwing a random subset of 2 lines)....these 2 lines I display (indexable) on product...and rest of the product description is viewable on click which opens in iFrame.
When you've got all of these similar products, can you identify that they are similar? Is there anything in the database that associates them all together?
1,000 pages with 2 random lines really isn't any good. What would be good is one page for each "product group" that has the full description, but then links out to individual product pages with the main product group page being the canonical url.
Actually the problem is that a particular supplier X posts same description is ALL the products, whether he is selling widgets or brownies.
I have canonicalized similar product... for example: except oldest red widget of a unique supplier x, all the later red widgets of supplier x are canonicalized. Similarly all green widgets are also canonicalized except oldest.
But other different products are also carrying the same description... for example: purple handkerchief is also carrying same description.
Whatever the products are, description is genereic like: please visit my website example.com, we are verified wholsalers..blah blah
It sounds like a lot of your User Generated Content is user generated garbage. Maybe you need to tighten up on Terms of Service and moderating contributions. Make contributors toe the line or you just don't publish them.
[edited by: tedster at 3:32 pm (utc) on Apr 6, 2011]