The 21st Century Indices
Important Information about Using the Indices
Please note, except for the digraph ch, the indices are in Czech alphabetical order. The reason for not completely following the Czech alphabetical sequence is that all English ch words would have ended up in the wrong place -- after h -- instead of being arranged among other c words. Consonants with diacritics file after the unadorned form of the consonant -- č follows c -- but vowels with and without diacritics are interfiled.
Only POFIS catalog numbers appear in the catalog index. The Society for Czechoslovak Philately is an excellent source for POFIS catalogs in the US. The Czechoslovak Specialist publishes from time to time tables aligning Scott, Michel, and POFIS numbering schemes.
Constructing the Indices.
A set of XSLT computer programs generated the web pages for both indices from a single XML input file. Because the indices are computer generated, only a few seconds are needed to add new entries, corrections, or changes to the indices.
Indexers face two major problems: (1) selecting the correct index term to represent a topic, and (2) selecting the correct form of that term. We acknowledge that the limited experience in both areas reflected by us in this endeavor will frustrate some knowledgeable users.
Indexers who use computer programs to generate indices face yet two more problems: (1) creating the appropriate input for the program, and (2) insuring the program produces the required output. The programming to produce these indices required learning four unfamiliar computer markup languages: XML, XML Schema, XSLT, and XPath. As our experience with these technologies matures, perhaps a more pleasing product will emerge.
In general, the process of index construction followed a somewhat recursive path. That is, a series of revisions of the initial XML Schema and XML document paralleled data input. At the same time, running XSLT transforms on the XML to produce HTML documents provided another stage for feedback. The design seems stable at this point (August 2008) but will likely undergo further development.
For the most part, only the main articles in The Specialist's Table of Contents fed the original data chain. With that task completed, each issue is now being revisited to pick up the tidbits concealed under generic headings: Letters to the Editor and that sort of thing.
Having exactly one data input point for each item undergoing the indexing process drove the design. The XSLT transforms permitted distribution of the data to any number of access points: authors, subjects, and POFIS catalog numbers. The notion of role -- creator, translator, editor and so on -- provided further access points for the author facet of the index.
The subjects portion presented far more design and implementation difficulties. The mechanical process of subdividing major headings took a surprisingly long period of time.
Choosing appropriate entries for the thesaurus presented still more taxing problems and those choices face constant revision. The initial effort, which attempted to align terms in the new index with those in the 20th Century Index has seen only partial success. For one thing, applying heading terms at the same level of granularity used in the 20th Century Index proved more than a match for the knowledge we had at hand. For another, the design for cross references, has been implemented only partially.
The hope is to implement some sort of public feedback facility to support ongoing additions and corrections That the indices can be rebuilt and distributed in less than ten seconds makes on the fly emendations feasible. Or, should no one care to contribute, the index still satisfies its author: after all, it is a personal index.