We explore ways in which to optimize the utility to the user of an electronic mathematics journal. We consider the following topics:
Examples in this discussion are taken from the two electronic math journals hosted at the University at Albany, the New York Journal of Mathematics and the Pacific Journal of Mathematics, Electronic Edition.
Prognostication: The rate of access of electronic mathematics journals from PCs will surpass the rate of access from unix systems in the near future.
Result: Access from a PC is likely to be more convenient and more responsive (in terms of the speed of client-side software) than access from a workstation.
(NYJM has received vociferous complaints from such users.)
Consequence: PDF and html (or more properly XML) based systems increase in value.
The latter are under development, and will take some years to reach the level of sophistication necessary for publishing electronic math journals. Potential drawbacks:
Developments here will be interesting.
Effect of inertia: Due to individual preferences, journals have found it necessary to accept submissions in a variety of TeX formats. Similarly, a variety of graphical formats of the published papers should be provided by the journals, regardless of the fact that some formats have better features than others. E.g.:
But this is due to more than inertia. For instance, users from a vanilla vt100 terminal may be getting postscript simply to print it. And users dialing up from PCs with TeX installations will opt for dvis, as they are the smallest format, and transfer quickly. Many users don't have a hypertex reader (which must be either bought or compiled, depending on the platform), and some dvi viewers may not be able to handle dvis which include the hypertex \specials. Thus, there is genuine demand among the readership for multiple formats.
Nevertheless, PDF seems the most versatile and valuable of the current formats:
Points in favor of PDF
Even in papers without graphics, pdf versions are significantly smaller than postscript versions of the same papers.
As such, it is unique in its ability to handle both text and graphics well. Any competing system would have to reinvent these capabilities. Thus, it has a viable niche, and is likely here for the long haul. Support from Adobe doesn't hurt, either.
Missing: a way to print pdf from a vt100 terminal.
Advantages and disadvantages of dvi versions
The advantage of dvis is that they do not contain the fonts at all, and hence are smaller than most other formats. But this is also a disadvantage, in that the reader must have a font installation (optimally also including the metafont program) to read or print them. Fontless pdfs are also possible (John Randall has produced demos: see http://math.albany.edu:8800/hm/emj/demos/Benson_fontless.pdf), and could play an analogous role if one could assume the reader had BSR or BaKoMa fonts installed on his/her platform.
The drawbacks of dvi posting concern the use of graphics (including diagrams) and hypertex links.
Larry Siebenmann has a program to remedy some of the above defects. Bill Hammond is considering others. Others are less tractable.
This is the next major problem to be faced in electronic publishing, in that the placement of such links should be as automatic as possible. We need a useful, standardized system of referencing. The problem is complicated by the existence of mirror sites.
It is also complicated by the existence of multiple formats of articles. Which format should receive the link? At NYJM and the Pacific Journal, each paper has an html abstract page with links to the various formats. That page functions as the paper's home page, and is the logical place to receive the link. But not all journals use this system.
People are currently experimenting with PII numbers for generating links. This is a long way from solving all of the above problems, but may help get a grid of links established between documents.
To be useful, the full TeX source of the articles should be indexed. At NYJM and Pacific, we then provide links to the html abstract pages of the articles containing a match. The actual TeX source need not be placed on the web.
We use Glimpse indexing. The output of the search displays lines in the documents that contain the keywords in question, permitting the user to determine the precise usage of the words in context. Thus, the reader interested in Hopf algebras can screen out the papers mentioning Hopf bundles or Wiener-Hopf operators from a search on "Hopf".
The next step is to develop a system for searching indices of multiple journals at different sites. This is currently under development among some of the independent journals.
These materials may include errata, comments, links to applications and references, computer programs that are used in the paper or that illustrate the paper's results, etc. See, e.g., http://nyjm.albany.edu:8000/j/v2/Zeilberger-info.html.
[S] M. Steinberger, Electronic Mathematics Journals, Notices of the American Mathematical Society 43 (1996), 13-16.