Registration -
Powered by Elgg

John Heap :: Blog :: Help! Quick! don't panic! Just do it!

May 31, 2007

default user icon
John Heap

Help! Quick! don't panic! Just do it!

Activity 7 already.  I don't remember the original email.  They (the powerful ones) will never believe I didn't receive it.  How do I get my community entry up?  Where?  When?  No time.  Too busy in Second Life!

Professor Janet Finlay (intimidating working with a real Prof) and I are the team looking at "Social Networking to Aid Resource Discovery" - believing that the failure of academic staff to engage with structured metadata might be ameliorated (is it OK to use big words in a blog .. and is that the right one?) by allowing them more sophisticated ways of 'phoning a trusted friend' to locate the resources they need.

Is that enough? It is for me.  I go on holiday on Saturday and I have no idea where my flip-flops are!

 


Overview for Keywords: community directory, Janet Finlay, John Heap, resource discovery, trusted friend

Blogs with Keywords: community directory, Janet Finlay, John Heap, resource discovery, trusted friend

Posted by John Heap

Comments

  1. Happy holidays, John. Structured metadata drives me mad because, of course, natural language is metadata for life and structured metadata is so not natural language. But "folksonomies" are so sloppy and all over the place: "blog", "blogs", "blogging" are all discrete entities. What is the answer? Postcard from paradise, please.

    George Robertsgeorge on Thursday, 31 May 2007, 23:19 UTC # |

  2. One of the fundamental issues is that the 'records' we work on are always incomplete and produced by people with different terms of reference, time, and inclination. The traditional metadata approach attempts to put quality control and policy on this to prevent 'bad metadata', but this necessarily restricts the amount of resources you have to work with.

    OTOH, I used to work for companies involved in intelligence analysis, where the data is also always incomplete and sloppy, and there a combination of latent semantic analysis, fuzzy matching and SoundEx was generally 'good enough' to support human analysts. I don't see why the same techniques can't improve on 'dumb tags' without creating any additional hassle for users and creators.

    Scott WilsonScott Wilson on Friday, 01 June 2007, 10:30 UTC # |

You must be logged in to post a comment.