David Mimno, Alison Jones, and Gregory Crane. "Finding a Catalog: Generating Analytical Catalog Records from Well Structured Digital Texts." Proceedings of the 2005 Joint Conference on Digital Libraries, Denver, CO, June 7-11, 2005.
On the whole, I found Mimno's article fascinating. The idea of automatic metadata generation intrigues me great, though I will admit to still being rather skeptical, even after reading this article.
One section that I found particularly interesting was when they were discussing the use of statistics and probability in the metadata generation. They use the example of Washington within the Civil War-Era documents and discuss how they use the placement, frequency, and words appearing before and after Washington to determine if it is the place (and if so, which Washington) or the person.
I thought that the article was presented very well. It was very understandable while discussing the various steps in the extraction of different types of information. I also found it helpful that they explained exactly which MODS fields the metadata would be entered into (e.g. the title area of the XML into titleInfo).
I also found their reasoning for choosing MODS was very sound. It works well within the constraints of traditional cataloging and mirrors tradition context more easily than Dublin Core, for example (an issue I have been getting to know well at my internship this summer). They also discussed the fact that MODS being in an XML format was also important. This makes MODS compatible will many different kinds of software for "editing, searching and formatting metadata records."
No comments:
Post a Comment