follow me on Twitter
    My Photo
    Location: Aberystwyth, Wales, United Kingdom

    I am the owner and managing director of Information Automation Limited (IAL), a company that specialises in research, consultancy and training for the information profession. We are particularly interested in all forms of electronic information resources (e-journals, e-books, etc) and I teach a course in electronic publishing at the Department of Information Studies in Aberystwyth. Drilling down still further(!), my interests centre on the quality and evaluation of electronic information, and in the thinking that underpins activities in library and information science.
    This is a Flickr badge showing public photos from chrisinwales. Make your own badge here.

    Tuesday, March 28, 2006

    Tagging is messy... but it works!

    The following is a quotation from Gary Price as reported in the Library Stuff blog, by Steven M. Cohen, published by Information Today, Inc. I'm not sure where they got it from but here it is:
    "Yes, tagging can be very powerful and useful for very individual or small focused groups like an eight grade class or a group of friends or co-workers. To be useful to the masses (if/when) it reaches mainstream/widespread is another matter. Is the point of tagging to make information retrieval more precise for a large group of users? Why? Synonyms, pluralization, etc. Also, spam and gaming the system. This is another topic NOT addressed in the article. What would stop someone reviewing the most popular tags and then including these tags in every item they post? I'm sure with several logins and a script this could be achieved quite easily. We all know what happened to the meta-keyword content tag. Aside from spamming, for tagging to save effort and make retrieval more precise, something I've called structured or fielded tagging (location field, author field, date field, etc.) is needed. The Catch-22 is that most people wouldn't do it. Others would say that in some cases on the open web, a document or other item should speak for itself and let technology like dynamic clustering, audio transcription, content-based image retrieval, etc., do the work for the masses. Again -- on a personal or small group level -- it's another matter. The same might also be true when it comes to small specialty or vertical databases. Librarians know that everyone isn't a cataloger. Btw, standardization amongst tags and tagging services is needed."
    I'm quite surprised at how many times these arguments are being rehearsed - perhaps the post on this blog that brings it all together is 'Social tagging or social decisions'. Obviously, indexing created by untrained indexers in a completely unstructured and uncontrolled way is never going to produce high precision or complete recall. Its messy! But some of the structures and mechanisms that surround tagging make it useful. For example, you discover a colleague working (tagging) in the same area of interest as yourself, and create an RSS feed so that you are alerted to anything s/he tags. It widens your horizons. It provides effortless retrieval at the cost of a few 'false drops'. And, as I have suggested before, it is especially powerful if used in conjunction with formal indexing/classification. For example, tagging in OPACs, which could let users tag items for level, relevance to work, value. The University of Huddersfield have been experimenting in this area.
    Standardization of tags would make it less messy, and some systems now offer suggestions to taggers in a kind of predictive typing way. Spamming the system is an obvious problem, but this is a social tool we are talking about - society (the taggers) will find a way to deal with it if it becomes necessary; society may even find the spamming useful!

    >>Technorati tags: ; ;

    >>IceRocket tags: ; ;


    Links to this post:

    Create a Link

    << Home