Monday, June 16, 2008

Tagmashes for Readers Advisory

I've been thinking a lot about how booksellers and librarians can use LibraryThing for "readers advisory," helping readers find books they'll love. One answer, I think, is to promote and improve our "tagmashes" feature.

Readers Advisory is something of a discipline in librarianship, with a body of thinking behind it. There are also a number of well-known subscription RA tools, such as NoveList and FictonConnection, available in a very large number of US libraries. (See this page for a much larger list, which includes LibraryThing up with the big guys.)

LibraryThing can be used for Readers Advisory in a couple of ways:
  • Some libraries have used LibraryThing to highlight special topics (eg., new YA material at the Framingham Library)
  • Most LibraryThing works include recommendations—both automatic and member suggested, and with various summary and detailed lists—so you can get from a known book to a set of similar titles.  
  • Our fielded wiki Common Knowledge links books by series, places, awards and so forth.
  • LibraryThing tag pages provide relevancy-ranked lists for many topics, eg., chick lit, steampunk, memetics, cozy mysteries
  • Tagmashes
"Tagmashes," introduced a year ago, are a variant on tags, for when a simple tag isn't good enough.

By combining two or more tags, or excluding tags,  tagmashes extend tagging and nip away at some of the unique values of traditional subject classification—high granularity and hierarchy. Thus, although the tagmash France, wwii doesn't have an explicit notion of hierarchy, it works something like the LCSH World War II, 1939-1945 -- France. (And, of course, the LCSH tree is an artificial one—there's nothing in the idea that makes France a branch of World War II more than World War II is a branch of France!)

Notably, the system doesn't make tagmashes, users do. Once made, they "stick around," and may appear on related tag and subject pages, with their overlap to that page listed, testimony that a particular combination of tags made sense to someone. The system could--but does not currently--track tagmashes for relevance and usage, pruning some and elevating others. And it could allow users to edit, rate or review them for useful and accuracy.

I have it in my head that tagmashes, particularly with these additions, are one stone in the bridge between "free tagging" and traditional classification, between algorithmic recommendations and hand-generated ones, between the physical past and the digital future.

I see a world of librarians and readers creating, spreading and editing book lists that don't just "stay still"—depreciating over time, like a physical object—but shift and grow like a digital object can. And they wouldn't be the same for everyone, like a physical object, but adapt to the reader, like only a digital object can.

Anyway, here are some tagmashes to play with:

Labels: , ,

7 Comments:

Anonymous Anonymous said...

If a user trusted your design to show them what information is important, or even what information exists in the database, they would never find tagmashes. It's not on search, item pages, etc. Maybe that's more the problem.

6/16/2008 5:04 PM  
Anonymous _Zoe_ said...

The system could--but does not currently--track tagmashes for relevance and usage, pruning some and elevating others. And it could allow users to edit, rate or review them for useful and accuracy.

Some sort of automated tagmash improvement might be interesting, but I'm really skeptical about the second part. Do you mean rating the pair/grouping of tags that comprises the tagmash or rating the list that the tagmash produces? I think the latter would do more harm than good. Tagmashes as they stand are an aggregate of many users' perceptions of books, so I don't see how emphasizing the opinions of certain users could make it better. "Romance, trash" brings up some Jean M. Auel books at the top. Should fans be able to say that that's wrong?

6/16/2008 5:25 PM  
Anonymous d@vid said...

Null sets (and low results) reveal another use for tagmashes: identifying obscure nooks. Perhaps just the genre crossover for an author to write in, or for a marketer (blegh) to identify growth potential.

For example, it looks like Poppy Z Brite currently dominates the zombie cooking market.

http://www.librarything.com/tag/cooking,zombies

I'm sure there's room for healthy competition there.

Agreed with anonymous that tagmashes could be more prominent. I see that they work if you enter comma-separated values in the tag search field, but perhaps a help link or hint would prompt more use?

6/16/2008 5:30 PM  
Anonymous hydrogenguy said...

"Romances involving zombies and taking part in Greece — Null set"

Tagmash has identified a need which must be met.

6/16/2008 8:36 PM  
Anonymous Donogh said...

I agree with anonymous#1's first point. There should be a bigger push on making these available to users (whether this is just by highlighting or promoting the tag-search, I don't know)
But this is an incredibly useful tool - a search for crusades, heresy brings me to a reading list for the Albigensian Crusade and the Cathars

6/19/2008 4:36 AM  
Anonymous jamist said...

d@vid's comment on cooking,zombies just begs the question(s):

Is it:

a). Cooking BY Zombies?
B). Cooking FOR Zombies?
C). Cooking OF Zombies?

(Ewww.)

6/23/2008 8:50 PM  
Blogger VictoriaPL said...

Even though Tagmash was introduced a year ago, I hadn't really worked with it until a few days ago. It's awesome. This is quickly becoming one of my favorite features of LT. Thanks Tim.

6/25/2008 3:28 PM  

Post a Comment

<< Home