Recall first the general goals and principles for the Colloquium services, including the four kinds of information that are maintained there:
Additional detail about these structures are found in [services]. We proceed now to the concrete organization of these structures in the computer systems.
The naming is done using mnemonic identifiers, in the customary fashion. We use simple and composite identifiers, where a composite identifier consists of two parts, each of which is a simple identifier. The following types of simple identifiers are used:
Dublication of identifiers possibly occurs if a certain article is first published in the archive, and then commented in the bibliography (possibly after it has been accepted and published in a conference or journal). It is a matter of choice whether one shall use single or double identifiers in such cases (to be discussed).
The systematic use of identifiers is of three kinds. They may be used as subdirectory names, as file names (always for .html files), and as anchors inside .html files. Recall the structure of host names and of access paths within the Colloquium structures, as described in [notyet.available].
For each of the simple identifier types, there is a corresponding directory file ("directory" in the informal sense, not as a directory within the file system), where the identifer is used as an anchor. For example, the directory of active researchers is an ordinary HTML file containing a few lines for each researcher, containing family name, first name, affiliation, E-mail address, and reference to his or her WWW home page. These lines are labelled with the person's identifier as an anchor. They also, additionally, contain links to other information about this person in the bibliography and repository. For example, it allows to link to the article list and article header for those articles of this author which are represented in the bibliography.
Conference occurrences, institute divisions, and network divisions are listed and explained in the directories for conferences, institutes, and networks, respectively.
For each bibliographic identifier e.g. Author94A, the BRS contains one header file whose name is /brs/Author/Author94A/index.html. Other information pertaining to that article is stored as other files in the same subdirectory. The author directory contains a file /brs/Author/index.html that effectively lists the currently indexed articles of that author.
Similarly, the EPA archive contains one header file for each entry. The report called lab-94a22 will have a header file called /epa/lab94/lab-94a22.html. This header file contains links to the full text of the article, possibly in several variants (e.g. one variant which looks good on the screen and another one which looks good on paper), as well as a separate abstract, a separate bibtex descriptor, and so on. All this other information will usually be at the author's home computer system, but the header files should be in a central place for stability reasons.
It is intended that bibliographic and archive header files shall be permanent, so that one can reliably make links to them from articles, reviews, etc. Lists of articles by a given authors, articles from a given node (laboratory), and so on, may however move around and change, and should not be referenced except from the main researcher directory and institute directory. Similarly, the links in the header files, such as the links to the physical location of the full article, is likely to change over time, so it is always recommended to link via the header file.
We presently investigate the possiblity of obtaining a separate Internet naming scheme for this bibliographic and repository information, so that it the whole structure can be transparently moved to another physical computer if necessary.