The index was prepared completely automatically from the BibTEX files for the journal, without any human intervention whatsoever.
Since natural language is often ambiguous and highly irregular, automated preparation of the index produces some deficiencies which are discussed below. Nevertheless, despite these shortcomings, the indexing software can be applied to scores of journal-specific bibliographies, thereby making reliable journal indexes readily and widely available, at very low cost and human effort.
Whereas hand preparation of a decennial journal index would likely require days of error-ridden, and extremely tedious, human labor, automated index production is totally consistent, and very much faster. On a modern UNIX workstation, the data for the decennial index for a journal can be extracted, indexed, and typeset in less than thirty seconds. Importantly, this means that as the basic bibliographic data is extended, improved, and corrected, the index can be regenerated immediately. Since most of the journal-specific bibliographies in the TEX User Group bibliography archive are updated many times a year, the indexes can be as well.
The final typeset form can be offered in a variety of forms, including TEX DVI, PDF, PostScript, and HTML for the World-Wide Web. Journal subscribers may find it convenient to print the index periodically, and include it with the journal issues on their shelves. Others may prefer to use the electronic forms of this document on the World-Wide Web.