[Sigia-l] Research papers on tagging

[Are Gulbrandsen]

On 2. apr. 2008, at 12.31, Paola Kathuria wrote:
> Stephen Collins wrote:
>> On another list, I'm having a discussion with an unbeliever
>> (ooooohhh!) about the value of tagging. His argument is that given
>> everything is searchable, everything automatically becomes a tag he
>> can use and tagging is therefore, noise.
>
> On Flickr, for example, I add tags to enhance findability. The
> tags don't duplicate information in the text.
>
> So, I'll add 'London' and 'England' (because there's more than one
> London) to a photo I took of a building, or 'flower' to the photos
> of my yellow peony.
>
> In the absence of a controlled taxonomy that people can choose
> from, tags are a way of providing the context that otherwise
> can't be extracted from just parsing the content.

This is a very interesting case where the content is poorly  
searchable without the social tagging. (This is of course a  
digression from Stephen Collins case of full text search augmented by  
social tagging)

The Library of Congress has made more than 3,000 photos available on  
its Flickr page, for which no copyright restrictions are known to exist.

They want people to tag, comment and make notes on the images, just  
like any other Flickr photo, and hope this will benefit not only the  
community but also the collections themselves. Many photos were  
missing key caption information such as where the photo was taken and  
who is pictured. If such information is collected via Flickr members,  
it can potentially enhance the quality of the bibliographic records  
for the images.

Here is an interesting example (borrowed from David Weinberger):
http://www.flickr.com/photos/library_of_congress/2179930812/in/ 
photostream/

People soon used up the available number of tags, and it doesn't seem  
to be a way for other people to vote on the existing tags.

I think the challenge is that people in general are very likely to  
associate any picture with a lot of subjects and anecdotes which are  
very subjective. The tags are meaningful within their own context and  
frame of reference, but will be noise for most other people.

I think the question is then whether the technical system used is  
able to harvest the wisdom of the crowd, or if all the associated  
tags and comments leads to too much noise and poorer findability.


Best Regards,
Are D. Gulbrandsen
The XML-group,
Center for Information Technology Services
University of Oslo