We've all been asked for books by colour: "I'm looking for an economics book. It was about this thick and it had a red cover". Because, obviously, we know what every single book looks like...
The Superpatron Ed Vielmetti points to Dave Pattern's work on presenting OPAC results based on the colour of the book. Between them, Ed and Dave have come up with some genius proto-search tools for colour-matching books in the catalog.
This is one of those fun, but potentially useful, things that the Web 2.0 paradigm ought to be able to offer us. It's a bit of window-dressing, but with what could be a seriously useful application. Play with it and tell me you don't think it's cool! What if you could enter your keyword in a search box and pick the colour of the book you want from a spectrum? What if you wanted to create a colour-themed display (which can be really effective) and have an online version via your OPAC?
Tuesday, 24 April 2007
"It was about this thick and it had a red cover"
Subscribe to:
Post Comments (Atom)
13 comments:
Well, I've said that RDA should have a colour component ;) And what of FRBR? Could that have [coverColouris] with an RGB value :D
FRBRised colour values - yes!! In all seriousness, I think it could be an interesting idea... Definitely useful if it's applied to spines, since that's what you see on the shelf.
This really made me laugh, I am very familiar with the "have you got that blue book?" enquiry and perhaps the best use for this tool (using the term loosley but with a degree of seriousness) would be to have an alternative OPAC for students to search for such elusive items themselves, perhaps they would then begin to understand that the problem, (not the answer), is in the question.
I'll have a think about other potential uses and get back to you.
Everybody knows that the big blue book is Haralambous on Sociology - don't they? Judging books by their covers indeed.
Seriously though, people are definitely giving this some careful thought. See here for getting the LT folksonomy into library catalogues. People there already tag books by location in their house for instance. The point is that it's meaningful to the person who wants to find the book to use it, unlike a lot of LCSH for instance. Aah, the eternal cataloguing dilemma of subject headings....
There's quite the debate on personal vs generic tags indeed. I mean how useful communally is 'Propping up the lounge table' or 'In place to shock my folks' as a tag ;) Such superpersonal tags are best kept hidden. I'm not sure, also, as to the usefulness of tags like 'boring' :D
As Dave and I are slowly discovering (not being color geeks either one of us) there's some subtlety to this color matching. A book that's half pure white and half pure black will show up as gray by some measures if you average things out (which would be wrong).
Even if it's wrong, though, you can probably get a good match for some queries by looking at mini-covers, or doing a subsequent sort by some measure of interestingness (# of recent checkouts, # of copies in circulation, # of books in Librarything).
"That green sociology book" is probably Durkheim - http://www.librarything.com/work/36084 matches the best for greenness and the longest time it's been that color on the shelf. I picked it on a search for the sociology tag + sort by most popular + view one cover at a time.
I've used Amazon in the past to tackle the dreaded book colour query, but with mixed success: Amazon's offerings far outweigh our holdings in terms of numbers. Cross-referencing with the user's loan history in the LMS helps, but it's still a bit hit-and-miss.
And I've seen very effective colour-themed displays: red for Valentine's Day, for example, although a bay of red books has been known to freak users out slightly. But whatever colour is picked - or even themed covers like landscapes, or aeroplanes, or people's faces - the displays are always striking.
Maybe moving toward a system that describes covers in more detail than "just" colour - although, frankly, that's probably a big enough job for Ed and Dave - would be even more mind-blowingly cool.
As for tagging, I've been spending a lot of time on that subject. It's probably worth a post on its own, and it would be really interesting to see some comments on it.
Incorporating LT's data into the OPAC would be interesting, and I'm sure it's come up on the NGC4LIB list. There has definitely been some discussion of the (lack of) value of personal tags
I wish my library had a Superpatron... Maybe s/he's out there somewhere!
Well, you could have a set-up for descriptions of images on covers etc.
Two of my MSc lecturers, Rafferty and Hidderley- they really should've been solving mysteries in a van with names like that- were into Democratic Indexing, which included a form of image tagging. Might be worth looking at their work.
I was discussing something similar with one of my colleagues the other day - basically an image search function which incorporated some tagging and could find other pictures of a similar composition - in terms of subject, colours etc. - via some sort of fuzzy logic kung fu. Or something.
I'll seek out your detective duo and see what I find. Thanks for the tip!
A pleasure. Seems Pauline and Rob are still working on DI, nowlooking at in relation to 'folksonomies, mob indexing and desire lines' according to her work site.
I wonder if anyone is working on the stupidity of crowds; the Pratchett Equation where CrowdIQ=LowestIQinCrowd ;)
Would those be the same crowds who voted for Robbie Willams as Songwriter of the Millenium? Is he *really* the best songwriter of the last 1000 years? Or is he in fact someone who benefitted from a talented co-writer and zeigeisty pop production techniques?
[Insert curmudgeonly-old-man rant about "the pop music these days" here]
Me, I'm a Zeppelin fan... Page, Plant, Bonham and Jones: now *there* are some songwriters of the millenium...
Well, and indeed, although I feel that Led Zep are hugely overrated. As are, oh, the seventies ;)
The whole DI argument relies on things tending to an acceptable point, but rather ignores the 'green pencil' issue; the people who contribute are motivated so to do and therefore don't always represent any sort of consensus, and often not even democracy ;)
There's some previous experiments with searching for covers here: www.daveyp.com/blog/.../172/
Post a Comment