On 27 Jun 2005, at 03:47, joshua schachter wrote:
Harmonizing URLs with sessions in them (like Amazon) I guess will
go some of the way towards fixing this. I could also trap URLs with
&isbn=..... and also look for urn:isbn:.... tags as well (which
people can use now.) I think in Amazon's case, ASINs for books ARE
the ISBN's.
But yeah, a "book" URL recognizer might be interesting.
I do exactly this on CiteULike, and turned out to be pretty easy to
implement. Some things I learned in the process are:
1) ISBNs are (generally) ASINs, but the converse does not hold.
Amazon sells, for example, microwave ovens and all sort of other
types of consumer electronics. For my purposes, I wanted to filter
these non-books out (it would look pretty strange seeing a
Playstation 2 posted to an academic citation service), but I daresay
you might potentially want to allow such things. It might even be a
business model for you to recommend "popular" and "related" things
you can buy. Users might trust this information more than anything
coming directly from the amazon.com web site. You never know.
2) The "generally" means that some books have multiple editions with
multiple ISBNs. So, in some cases, a book will have different ISBNs
in different countries. Other than doing string matching on the
title, and authors, I'm not sure how to map international editions of
the same book down to each other (or whether that's even desirable).
The key point though is that only some of the Amazons
(.co.uk, .com, .fr, etc) will sell a particular book with a
particular ASIN, so you can't do the obvious thing of just rewriting
the link to the user's local Amazon and hoping the book will exist.
Furthermore, much as they'd like to, they don't sell every book in
the world. If you discover an obscure ISBN from some other source,
you'll just get a (prettily formatted) 404 if you link it to Amazon
and they don't have it.
3) The Amazon API is excellent for extracting the details of the book
(Title, Author, Publisher, etc) and capturing those in your own
database. It also solves the problem in 2 as you can (gently) submit
API requests to each of the international stores to see if it exists
on a particular site.
4) They have painstakingly scanned the covers of all the books and
you can really help navigation on your site by using them (E.g.
<http://www.citeulike.org/tag/economics>). Moreover, you can serve
the images from Amazon's serves: it doesn't cost you anything in
bandwidth, and Amazon are only too happy to let you as they
anticipate making a sale out of the deal.
Richard.
_______________________________________________
discuss mailing list
discuss@del.icio.us
http://lists.del.icio.us/cgi-bin/mailman/listinfo/discuss
