Functionality / design requirements

Metadata augmentation and performance

How can we get metadata augmentation from many different sources in a way that doesn’t create gatekeepers but also permits high performance searches?

Metadata augmentation would permit classification or other information about a resource to be searchable from any node on a network, regardless of who publishes the resource. In traditional metadata models, only the publisher of a resource can provide metadata and search capability about the resource.


Submitted by

Stage: Active

Feedback Score

0 votes

Idea Details

Similar Ideas [ 4 ]


  1. Comment

    Unless you plan to extract the data you want from free form descriptive contributions there will always be gatekeepers. The design of the form for me submit information is a gatekeeper of sorts. The standard approach to free form text is indexing which is partly useful. If you first do some semantic analysis of content, which is rarer but more valuable, then your indexing will be better and your efficiency will improve. It all depends, however, on how your end users actually try to find what they want. This will vary dramatically based on technical ability, domain of interest, community of interest etc. Successful interaction is based on successful anticipation of user behavior.

    There may be some more important questions in relation to gatekeepers:

    1. Are gatekeepers always a bad thing?

    2. Are there ways community-based gatekeepers can help?

    3. When do you want to avoid gatekeeping?

    Also, there is likely to be value in strategies other than augmentation. What about cascading metadata based on taxonomies that are understood by a particular community? That would, of course, require community-based flexibility with taxonomies. Perhaps some form of gatekeeping might assist in that approach. We know that taxonomies are not the perfect answer for discoverability but that does not mean they should be ignored where they can provide value.

    I think the core requirements of many of the issues will go back to understanding the user communities quite deeply. That would include your solutions for augmentation, federation, flexibility, scalability, classification etc etc.

  2. Comment
    steve.midgley ( Idea Submitter )

    Fair point allynr. When we're talking about gatekeepers it's not in the context that *all* "technical" gatekeeping is a problem. That's like saying we want a WWW with no webservers.. :)

    With regard for metadata augmentation, what we're investigating is how to let metadata be contributed by anyone and the *only* gatekeeper to their contribution is the "metadata server" they use to push the information into the network. So they don't have to ask permission from some other organization to put their info into the network (like the org who posted the resource they are augmenting with new metadata).

    I'm not sure I follow along about the cascading metadata concept. We're all in favor of taxonomy/ontologies/folksonomies to describe resources, but the way I see that's just another piece of metadata (though possibly highly structured). Are we missing something in that light?

    Agree that user communities matter: in particular we've been trying to figure out how to connect to user communities rather than directly with users (as much). If we can provide some data services that are valuable to orgs that are already "curating" communities, that seems a lot easier "lift" than trying to build something for everyone?

    Thanks for you thoughts -- feel free to post back of course.

  3. Comment

    Hi Steve

    I guess the comment I have posted in the solution design area is quite relevant here and also in the other metadata discussion. I won't repeat it here.

    I agree with all the points you made in your reply and really like the thinking that is going into this project.

    The cascading metadata approach is one that can be used to handle inferred metadata for content that has less or different metadata than you may need to facilitate discovery. It is an approach that can be used in repositories where the nested categories assume the metadata of parent categories to be relevant. If the initial contributors to the registry all have the required metadata then this is less of an issue, however, when providing metadata to different communities then some cascading inference might be useful. Just a thought.

    From all I can see so far, the communities/existing services etc are going to play a very important role in the way that you conceptualize the solution and the requirements.

    Thanks for your reply it was helpful to me as well.


Add your comment