Amazon and the Library of Congress: Together at last
Well, I think I've come up with the right solution, using Amazon and the LC together. It's a little complicated, but the complexity is hidden from the user.
It works something like this. First it looks in the LC. If it can't find it there—either because it's not there or because the search didn't follow LC rules—it goes to Amazon. If it finds it on Amazon, it makes one last heroic and generally successful effort to find it at the LC, this time using Amazon's data in a LC lookup.
People who don't care about LC data can structure their search as loosely as they want and will still end up with LC data most of the time. People who must have LC results can make sure they get them. If you have any doubt, the Add books screen tells you where the data came from for each book.
When there's LC data, it tends to prefer it over Amazon data. This is because Amazon plays a bit loose with authors and titles. Authors are first-last sometimes, last-first others. Titles often include the name of the series the book belongs too. The LC is more careful. At the same time, it always uses Amazon date and publication info. This ensures that, although the LC may have an older edition, your info will match the book you clicked.
Inevitably the multiple sources hamper attempts to "match up" equivalent books. Right now it tends to match books up by LC control number (which can embrace two ISBNs) or by ISBN. In the future I'll be doing a more sophisticated sameness test, involving titles, authors and other data. The same/different issue can never be solved fully, but I'll try to strike a reasonable balance.
Confused? Don't be. I think it works pretty well. Feel free to differ.
No response from Amazon yet. If they insist on freezing data and requiring constant refreshes, I will have to make some changes.
It works something like this. First it looks in the LC. If it can't find it there—either because it's not there or because the search didn't follow LC rules—it goes to Amazon. If it finds it on Amazon, it makes one last heroic and generally successful effort to find it at the LC, this time using Amazon's data in a LC lookup.
People who don't care about LC data can structure their search as loosely as they want and will still end up with LC data most of the time. People who must have LC results can make sure they get them. If you have any doubt, the Add books screen tells you where the data came from for each book.
When there's LC data, it tends to prefer it over Amazon data. This is because Amazon plays a bit loose with authors and titles. Authors are first-last sometimes, last-first others. Titles often include the name of the series the book belongs too. The LC is more careful. At the same time, it always uses Amazon date and publication info. This ensures that, although the LC may have an older edition, your info will match the book you clicked.
Inevitably the multiple sources hamper attempts to "match up" equivalent books. Right now it tends to match books up by LC control number (which can embrace two ISBNs) or by ISBN. In the future I'll be doing a more sophisticated sameness test, involving titles, authors and other data. The same/different issue can never be solved fully, but I'll try to strike a reasonable balance.
Confused? Don't be. I think it works pretty well. Feel free to differ.
No response from Amazon yet. If they insist on freezing data and requiring constant refreshes, I will have to make some changes.
posted by Tim at 4:21 AM
7 Comments:
As I understand this, you're still only pulling the records from Library of Congress so that there are still no records available for all the books (mostly genre fiction and especially fantasy) where the Library of Congress doesn't have a copy of the book. It seems like at least 50% of my collection is not held at the Library of Congress. Wouldn't an easy fix be to have the secondary search run a Z39.50 search against the catalogs one of the large Public Libraries (Chicago, LA or NY) or mixed library-type consortia (e.g. MELCat in Michigan)?
Oh--Now I've figured out how to force records from Amazon. I have to be sloppy in my searches. Do you know how difficult that is for a librarian?
Is there a way to put a location into a field? E.g. "bedroom" "garage" etc.?
I'm having trouble getting the database to recognize the right version of some books: specifically, I have the UK editions of all the Harry Potter books (I'm a snob that way) and I can only get the searches to find the US versions. Do I just need to enter mine manually, and forego the pretty cover art?
I'm installing Amazon UK support as we speak, actually. Give it maybe an hour at most. There will be a drop-down menu, I think on the Add books page.
Is there a problem with the Amazon searches? I've tried 0-09-947209-0 and 0099472090, both of which hit on amazon.com and amazon.co.uk but return nothing here, even if I ask for the Amazon sites to be searched first. I'd be here with my $10 if I could get searching to work reliably.
It would be cool if LibraryThing could fetch metadata for books published outside the US. Many countries (eg. Sweden where I live) have a national library which keeps track of all published works. In most cases you can get a record in MARC-format.
You can then program LT to fetch data from different sources depending on the ISBN-number searched for (91xxxxxxxx is Swedish publications for example).
It would be extremly helphul!
Post a Comment
<< Home