This interesting paper deals with the use of metadata to classify the contents of a digital library for astronomy, containing images, papers, tables, and even people. I would worry that the cataloging effort involved here may be prohibitive for many situations. The paper explains why one wants metadata sufficiently detailed to do searches on sky location, spectral position, date, and so on, but does not give detail on what the cataloging to make this possible will cost. Right now, indexing Web pages is extremely expensive; compared with cataloging a book. There aren't a lot of standards for the cataloging formats, you don't conveniently find much of the information you need on the back of the title page, and unlike a book you have to check regularly to see that the content is still what it was when you wrote down the metadata. The definition of AML (astronomy macro language) helps with the lack of standards, but the paper could really use some data on how practical the metadata schema is to implement, and how effectively the users can employ it. The users apparently can both browse and search the structure; it will be interesting to see how much of each they do, and whether they are able to make effective use of Z39.50 searching. The paper could also use some basic information about how big the database is now and what the eventual size will be.
a service of  Schloss Dagstuhl - Leibniz Center for Informatics