Users are used to changing their own information, my space, amazon etc. Can't even change your own address in the typical library catalog.
Face it, LCSH is not the most intuitive way to search for items. i.e. Motion Pictures. No way to describe an Action Movie. It is impossible for controlled vocabulary to keep up.
Folksonomy. Tag it in a way to make information useful to them.
Furl & Del.icio.us Store bookmarks on a webpage. Can use the popular tas that other people who have used it choose.
There is a remarkable concensus in the tags that people use. Lots of sharing. Less chaotic that you think at first as people go back and join the most popular shared language for their own tags.
Create online community of interests. Create tags, public tags could be shared across other users of the catalog. See what people are reading to create readers advisory targeted.
She's looking at the structure of the tags. Furl Del.icio.us and Technorati
In the future she wants to look at tagging behavior. Why are you choosing what you are choosing. What makes you go back to change them?
OCLC funded the research and thus acknowledge's that LCSH may not be the be-all-and-end-all.
Georgia Tech trying to create and alternative interface to the opac exporting allt he bib records as marcxml.
Move back and forth between folksonomic interfances and LCSH so you can use other folksonomic interfaces. (He hasn't done anything towards that in the catalog ;)
Enabling tagging isn't difficult technically. but your OPAc isn't a huge sell. What have you got that a larger community doesn't.
Just because you build it doesn't guarantee anybody actually comking.
18,000 FT enrollment. What % of that enrollment would use this thing? Niche stuff here.... Catalog isn't going to light up the undergraduates. No fiction catalog. Academics probably aren't the best ones to build the community/social software into.
In the meantime we've seen. CiteULike Tale Links PennTags Connotea Library Thing Amazon Tags.
CiteULIke bookmark and tag peer-reviewed scholarly journal articles.
Can be added via bookmarket for certain publishers/aggregators or uploaded from BibTex/EndNote/ReWorks/etc.
Active community 1000 posts a day
Support for grouns
based on unalog
little to no institutional promotion.
No real integration with library resources (OPAC)
Service will most likely be ended soon.
People outside of Yale couldn't access.
Seemingly strong institutional supoprt
Relativelt small (yet active)
userbase 425 (20,000FTE)
Tags and annotations appear in OPAC
given size of community and catalog there are few and far between associations between tags and items in collection.
Created by Nature Publishing Group
Similar to CiteULike in scope/functioanlisty except no group functionality.
Similar numbers to CiteUlike
Open URL support
Allows users to catalog their personal collections. (He doesn't udnerstand the attractions)
3M records 46.5K users
Grabs metadata from Amazon, LOC etc.
Recently added tagging
Gives Amazon yet another avenue to hawk their wares to consumers
Tags quality wildly vary editorial comments, etc...
Sole purpose seems to be to promote more sales.
Implementing our own taggins system doesn't make much sense right now. Want to be able to relate tags to controled vocabulary. Working on going that route.
1. Social bookmarking support added to the umlaut and Open URL Link Reseolveer serivces.
Sits over the SFX local catalog, consortial catalog, Amazon, OAI providers and stores any subject descriptions glenaed from sources.
(Its a little slow right now)
Stored subject term and where it came from)
Check the fulltext target urls against Connotea, unalog and Yahoo's MyWeb2.0 to see if anyone has marked these things (agged) remotely.
Store tags and tagcount.
Store a ball of subjects around a specific citation.
All terms given same weight LCSH headings and subjeadings get dividced up.
We don't care where they came from.
More matching tokens =greater relevance. Recommender service.
No browsing on a single term
Weed out the "me" tag (Flickr statistically irrelevant)
Hopefully this will create a map between LCSH MESH OAI subject Keywords/SFX Categories/tags to guess appropriate subjects of r citations without matchies in all areas.
Identify which term set yer coming from and bring you back to the ball of subject terms ont he web. Find things that might be releated to this things on the web based on other related things that we do know about.
URLS are very weak identifiers of items.
"Problem" Items are only tagged when they are found useful and wish to be fund again -if no one has found it how to you tag it to make it more useful. Getting out the great stuff that no one is using.
Wuality and focus of the source ins key
The larger the data/user pool the more useful the data.
Datamining project by Rob Sanderson MELVYL exported 18M records.
Found 127,896 terms in the 650a
only 47,319 appear more than 5 times.
40% being used just once. We're not particularily good with our own taxonomies. Lots of work to be done on both sides of the street.
Between the controlled vocab and the tagging we should be able to sync it up with we have known knows. Colloborate with sites that will work with our missions. Librarything CiteULike and Connotea could aid buildigng vocabyularies with identifieable datea.
Journal recores tag file =all of the article tags?
Authority maky take the form of associating the proper community to a given objects.
My Space Flickr etc. would give you too hight a noise to
MARS is the authority vendor
Yale creates all records in RLIN (that will have folksonomy)
While records are at MARS we use a local method ot "lock" them. It warns a catalog that its away for processing.
Load the processed records and overlay them over the voyager catalog.
Japanesse Arabic Chienese Korean Polish? Hebrew Yiddish
MARS is the Backstage Library Works people. Currently they don't touch the vernacular headings. Don't have the resources to do that atm. They are looking at it for the fugure.
They print out the romanized changes for Yale so that they have the ability to go in and make modifications.
"Normalized data" take out diacritics and extraneousespaces and do the same with the LC data and try to make it match. Cant do that with the CJJK characters due to certain diacritics in the name. Could be a totally different name.