Playing tag
Tim at the Librarything recently posted the results of study he did on tagging. As I read this it made me think about some of the points in Tim O'Reilly's article What is Web2.0 (I highly reccomend the O'Reilly article btw).
1. If tagging is ultimately going to be useful then it needs to improve search. We need to find a way to factor user generated content into our search algorithms, hopefully in a way that the user will come to understand that tagging or other content creation activities will improve their ability to find what they are looking for.
2. In order to reap web2.0 benefits (the benefits of operating within a network) we need to find ways to enhance data that go beyond asking users to consciously tag or categorize data for us. In other words we need to be documenting what users are doing and use what we learn to improve our data. Probably the paradigmatic instance of this is Amazon's users who bought x also bought a, b, c etc.
So thinking about this in the library world... I wonder what if circulation stats were figured into keyword (and heading searches) so that books that circulated more often were more likely to float to the top of results lists. What about tracking what books are put on reserve? Or what books do we own multiple copies of, or hold request and so on. Could we assume that a book that has been on reserve is more likely to interest a student than one that hasn't? Also could we think about generating tags from things that are already happening with books, for instance statistical codes we apply, funds, reserve lists, ref. works folders...
Finally I think Tim Spalding (Librarything) has a good point that tags are only (or especially) useful in large numbers. Thus a consortia catalog is much more likely to generate helpful tagging then any single institution. O'Reilly argues that true web2.0 applications perform better the more people use them. All this makes me think that the future of "SOPACs" lies with big conglomerate catalogs like IShare, WorldCat, or Librarything.
1. If tagging is ultimately going to be useful then it needs to improve search. We need to find a way to factor user generated content into our search algorithms, hopefully in a way that the user will come to understand that tagging or other content creation activities will improve their ability to find what they are looking for.
2. In order to reap web2.0 benefits (the benefits of operating within a network) we need to find ways to enhance data that go beyond asking users to consciously tag or categorize data for us. In other words we need to be documenting what users are doing and use what we learn to improve our data. Probably the paradigmatic instance of this is Amazon's users who bought x also bought a, b, c etc.
So thinking about this in the library world... I wonder what if circulation stats were figured into keyword (and heading searches) so that books that circulated more often were more likely to float to the top of results lists. What about tracking what books are put on reserve? Or what books do we own multiple copies of, or hold request and so on. Could we assume that a book that has been on reserve is more likely to interest a student than one that hasn't? Also could we think about generating tags from things that are already happening with books, for instance statistical codes we apply, funds, reserve lists, ref. works folders...
Finally I think Tim Spalding (Librarything) has a good point that tags are only (or especially) useful in large numbers. Thus a consortia catalog is much more likely to generate helpful tagging then any single institution. O'Reilly argues that true web2.0 applications perform better the more people use them. All this makes me think that the future of "SOPACs" lies with big conglomerate catalogs like IShare, WorldCat, or Librarything.
posted by Matt @ 12:29 PM
3 Comments:
At 10:30 AM,
Everett said…
Two thoughts relative to this without having yet read the original post.
If there is no way to get circ. stats factored into worldcat searching, as seems likely, how much risk do we run of fostering a sort of group think focused on newest and most popular materials to the exclusion of older or more obscure works? Chris Anderson noted instances in The Long Tail where this kind of social linking to songs on a music website eventually became circular.
Possibly contrary to what I just said rather than at the WorldCat or IShare level I think something like this could be very helpful if we could work it out within ACTS or if there were a way to limit by type of library
At 12:02 PM,
Matt said…
I think Everett you have hit on one of the big issues of the "Google" era. Can popularity really measure quality or does it just measure popularity? When does using popularity become so circular so as to be a barrier rather than an aid?
I think these are good questions. But I wonder what the alternatives are. In Trincat if I do a keyword search on Scotland I get pages of hits apparently based on where and how often Scotland appears in the record. At least 80 of these received the highest 5 bar relevancy ranking. I not sure how it decides which records go on the first few pages but I think it would be helpful if it factored in use and/or if you could resort by use.
Still this is worth thinking about more...
At 12:11 PM,
Matt said…
Comment 2
I think if our library software ought to be capable of sharing information with other systems. In other words we shouldn't think of integrated library systems of just letting circ. talk to acq. and cataloging. I think that if there were more shared standards for storing library data this should be possible.
I could imagine a WorldCat where you could select to search just the ATLA or ACTS schools and see results influenced only by that data. I think that would be really interesting and at least help novice searchers (and collection development staff) realize import works to be familiar with or add to our collection.
Post a Comment
<< Home