Admittedly I found it difficult to weed through the jargon here, but think I grasped the big picture. I can't help but consider the fragility of any database, but at the same time don't think that's an appropriate characterization. Given the immense nature of databases, I'm interested in how we negotiate an appropriate amount of error. Do any institutional recommendations for margin of error exist - or, maybe first, what auditing strategies exist for this purpose?
2) Introduction to Metadata by Anne J. Gilliland
In the conclusion her article, Gilliland emphasizes "carefully designed" metadata as being the important building block of information before offering suggestions for how to develop a good metadata base. Aside from the issue of time and resources, I wonder what other perils might result from over doing data tagging? In other words, does "careful design" imply a necessary balance needed between the amount of metadata for an object and the quality or precision of that metadata?
3) An overview of the Dublin Core Data Model by Eric J. Miller
Well, I should look into completing all the reading before posting ... That said, I appreciate this article for giving an overview of the difficulties arriving at One True Standard - which, of course, does not and will never exist. This writing dates back to 1999, I'm curious how effective it has been worldwide.
Posts
4 comments:
Your comment on the "Introduction to Metadata" highlighted an idea I hadn't noticed. I suppose too much metadata could be added to an item. I wonder what that would do to the process of finding the work. I know with MARC, even though that is standard metadata, if you do a search in a union catalog, works that are the same will show up under different hits because they were entered slightly different.
I found the database article to be way over my head too.
And I'm curious if the author went back and corrected his typos?
I also wonder what you think of as perilous when over tagging an item?
Perhaps rjz would like to help proof my postings before they are published ...
Regarding the question of over tagging, I was thinking about the importance of specificity - and thinking of the process more abstractly. At the extremes, too many or too general terms may make a record too exposed while too few or too specific terms may bury a record. I'm interested in exploring a balance in this continuum and would like to know what strategies could help achieve that balance. Also, what flaws exist in any established industry standards?
Maybe the flaw they are overlooking is that they are DCMI are tagging such an immense amount of information, seemingly not taking quality of information into account. And about the human error factor, I believe there is a 'human error' percentage they factor in when designing systems.
Post a Comment