Metadata for Electronic Journals in the Electronic Library of Mathematics

Table of Contents

1. Executive Summary

2. Introduction

ELibM offers a variety of services to journal editors as well as to the prospective readers. Among these are the production of electronic versions of mathematical journals and the mirroring of electronic journals which are installed on a web site already. Journals produced or mirrored by ELibM are disseminated through the EMIS mirroring system comprising more than 30 mirror sites worldwide.

Prior to the start of EULER, there were no metadata collected or supplied by ELibM in the form of retrievable databases. Consequently, the task of providing such metadata was different than it was for the participating libraries with their already existing OPACs or for bibliographic information services with their well-defined fieldings. ELibM had to develop from scratch such a database along with mechanisms to ensure the collection and updating of such information.

The original task description called only for a prototypical extraction of metadata for a selected subset of electronic journals carried in ELibM. This subset was selected among the electronic journals produced exclusively by ELibM, for which metadata could be extracted from the content description files provided by the publishers.

The task was subsequently extended to cover the whole of ELibM by incorporating and suitably adapting metadata from other information providers. Complete coverage of ELibM has not been achieved completely by the time of publication of this report, but is planned be completed before the project ends (with some restrictions on field contents, see Section 6).

It turned out that the journal production process required a reorganisation and reimplementation to allow for the gathering of metadata for EULER. The difficulties of this effort had been underestimated in the original work plan, a fact which commanded the shifting of significant efforts (from 1 MM to 7 MM) to this task.

3. Description of the ELibM collection

ElibM carries currently the full texts of 36 electronic journals (plus 2 discontinued journals which are still archived and will be indexed as well).

The greater part, namely 26, of the journals offered through ELibM, have an own web site which is mirrored by ElibM, in most cases without alteration. These will be called "mirrored journals" throughout this document.

The remaining 10 journals have their web pages produced or hosted by ElibM ("ELibM-hosted journals"). For 2 of these, the publishers do provide the complete HTML pages for the journals, thus ELibM is only the host, but does not produce the pages themselves. The remaining 8 journals have their web pages completely produced by ELibM and will be referred to as "ElibM-produced journals" throughout this document.

There has been considerable growth of ElibM both in terms of the numbers of journals and of available full texts. In the course of the EULER project alone, 7 new journals were added to ELibM. Another 8 journals are under preparation at the time of writing, 4 of which are envisioned to be added to ElibM before the end of the EULER project. All of these 4 new journals will be produced by ELibM. Of the 11 additions during the EULER project, 6 are produced by and hosted exclusively by ELibM, pointing to the growing acceptance of the production offerings of ELibM.

For ElibM-produced journals, the journal editors provide ElibM with content description information in a particular format along with TeX, DVI, PostScript, and/or PDF files of the articles. ElibM then creates the web pages from this information, making use of a production system developed in-house for this specific purpose.

Apart from providing production and distribution of electronic journals, ElibM also offers the same services for electronic article collections (such as conference proceedings) and electronic monographs.

ELibM Statistics (as of 25 April 2000):

4. Description of Method Used / Work Done

The conversion was done by adding to the ElibM-production programs the specific functionality to create EULER-compliant resource description files suitable to be fed into a Z39.50 database with the tools provided by CWI and NetLab for this purpose. Due to the specific design of ELibM-production scripts, the resource description files contain all the records of one specific journal issue. This means, in particular, that the several files/record option in the Zebra configuration file had to be activated for creating an EULER-compliant database.

The first installation of the DC-based frontend Z39.50 database with the test data set was made public in October 4, 1999 and incorporated into the Alpha-version of EULER immediately afterwards. In the verification phase, some problems, mainly concerning unconverted TeX encodings present in the records were addressed.

Another essentail part of the task was the revision of the technical guidelines for journal preparation. The revised guidelines now also contain material on metadata pertaining to mirrored journals (see Appendix 3). This makes possible for journal editors to cooperate in providing metadata within their journal webpages or separately. Appopriate procedures to gather metadata provided in these ways were defined.

Furthermore, in cooperation with FIZ Karlsruhe, which cares for the technical administration of EMIS, procedures for the exchange of metadata were defined. Specifically, FIZ provided metadata of 23 mirrored journals in ELibM. These metadata are currently adapted for the inclusion in EULER.

5. Analysis and Findings

Complete metadata records could still be produced for journals without or with incomplete content description files by adapting the metadata routinely collected by FIZ to the specific purposes of ELibM. For this, the FIZ records had to be converted and adapted to ELibM naming schemes and the EULER DC requirements. Special care has to be taken of the delivery information contained in the EULER-DC fields <IDL> and <OI> (see the EULER-DC specifications for definitions). The data provided by FIZ have such a field, but for mirrored journals it often points at the original web site rather than the location within ELibM, and also often contains only the URL of just the journal's home page or of contents pages, not the specific, exact document location(s).

In this context, it is also necessary to define what is meant by "document location". See, for example, http://www.emis.de/journals/DMJDMV/vol-05/vol-05.html for a typical contents page of a (mirrored) journal. Each article provided there has links to an abstract as well as to 3 full text files in different formats for downloading. The abstract is itself an ASCII or HTML page that may (but need not) contain links to the downloadable full texts.

It would be ideal to provide for all documents hosted by ELibM a separate page which lists all download options, or to link to the (mirrored) journal-provided abstract page if it contains such links. Such an "article information page" (meaning a page containing all metadata including links to full text files) is available for all ELibM-produced journals which are made from content description files.

For those journals, however, for which there are no such "article information pages", there are the possibilities of either producing one from the metadata available or of linking to all full texts directly from the EULER <IDL> and <OI> fields. Since there would be several links, the fields would have to be repeated several times. To display the size of the respective downloads (which matters for many users) would somewhat violate the field specifications and could pose compatibility problems with the other partners and with the engine interface. For these reasons, the first possibility is favored, i.e., preparation of a single article information page within EMIS.

To have suitable article information pages is, of course, a design decision the publisher can make. Therefore, the ELibM journal preparation guidelines were revised to include provisions for mirrored journals aiming at helping them with the design decisions and defining two procedures for providing resp. identifying the metadata themselves along with such pages.

The first procedure requires the journal to provide (X)HTML-embedded DC-based metadata within the document heads of the page describing a specific document. If no such (X)HTML pages exist (e.g., only one page describing the content of a whole issue), the journals are advised to readjust the structure of their webpages accordingly.  The metadata provided in such a way are then extracted with proprietary scripts that were specifically prepared for this purpose. This procedure requires the journal editors to utilize some form of metadata creation tool; for this purpose, the EULER Dublin Core Metadata Template provided by Netlab will be proposed to the journals.

The second procedure utilizes templates for the abstract pages which describe a specific document. In these templates, all variable information, in particular the metadata pertaining to the document described, are replaced by field name indicators structured according to the metadata scheme adopted by EULER. Accordingly, a template is an HTML document which contains the designations of included information and not the information itself. For example, the field name indicator "@authors" is used to designate the place in the abstract page where the authors of the described document are listed. All these fields are repeatable. Other variable information which does not fit into the metadata scheme of EULER carries the field indicator "@other".

6. Unresolved points

For mirrored journals for which there are no abstract pages which contain links to the downloadable full texts, the problem of obtaining a correspondence ("matching") between the files present in EMIS and the article metadata obtained via FIZ remains. Matching these two pieces of readily available information can in principle be automatized by applying templates as described in the Technical Guidelines.

It is safe to assume that there will remain a significant fraction of mirrored journals whose publishers will neither provide the metadata themselves nor adhere to the strict design limitations imposed by the template method. For those journals, the matching will either have to be done by hand, or more sophisticated pattern matching routines will have to be developed for this purpose.

7. Conclusions

Considerable efforts had to be taken to ensure that high-quality metadata can be collected for the heterogeneus journal base included in ElibM. As the process had to be initiated from the very beginning, it was difficult to foresee the timeframe in which the adaptations could be implemented. It turned out that the original allocations for this task were underestimated and had to be corrected accordingly during the process.

The production and collection of metadata for ELibM-hosted electronic documents that is undertaken within the EULER project will allow for the inclusion of new services in the future, such as automated transfer of bibliographic data to bibliographical information services and to alerting services.

8. Recommendations

9. References

A. Annexes

A.1 List of Categories kept in the ELibM Metadata

Alternative title
DC.Title.Alternative - TIA
not used in ElibM (may change in the future)

Corporate Author
DC.Creator.CorporateName - CA
not used in ElibM

Corporate Contributor
DC.Contributor.CorporateName - COC
not used in ElibM

Library of Congress Subject Headings (LCSH) keyword
DC.Subject - SUL
not used in ElibM

Dewey Decimal Classification (DDC) classification
DC.Subject - SUD
not used in ElibM

Computing Classification System
DC.Subject - SUC
not used in ElibM

URN
DC.Identifier - IDN
not used in ElibM (will most probably be implemented in the future)

Full text
EULER.Fulltext - FT
not used in ElibM

Record creator
EULER.Record.Creator - RC
not used in ElibM

A.2 Examples of Items from the ELibM EULER Database

There is exactly one sample document record for each journal for which content description files have been provided by the publishers (2 of these journals are in a preparatory stage before inclusion in ELibM).

<XREC>
<IDF>Acta Mathematica Academiae Paedagogicae Nyíregyháziensis, Vol. 16, pp. 15-24</IDF>
<DA>2000</DA>
<IDS>0866-0182</IDS>
<CR>Túri, Árpád Száz and József</CR>
<TI>Seminorm generating relations and  their Minkowski functionals</TI>
<LA>en</LA>
<SU>Absorbing</SU>
<SU>balanced</SU>
<SU>convex sets; balanced valued linear relations; Minkowski functionals of sets and relations</SU>
<DE>We show that instead of the Minkowski functionals of absorbing, balanced, convex subsets of a vector space $X$ it is more convenient to consider first the Minkowski functionals of balanced valued linear relations of $\0\Bbb R_{+}$ onto X$.</DE>
<RS>ELibEMS:/16//amapn16_3</RS>
<DID>/16//amapn16_3</DID>
<OI>http://www.emis.de/journals/AMAPN/vol16/3.html</OI>
</XREC>

<XREC>
<IDF>Balkan Journal of Geometry and Its Applications (BJGA), Vol. 4, No. 2, pp. 1-8</IDF>
<DA>1999</DA>
<IDS>1224-2780</IDS>
<CR>Neto, J.X. da Cruz</CR>
<CR>Ferreira, O.P.</CR>
<CR>Perez, L.R.Lucambio</CR>
<TI>A Proximal Regularization of the Steepest Descent Method in Riemannian Manifold</TI>
<LA>en</LA>
<SUM>49M10</SUM>
<SUM>90C30</SUM>
<SUM>53C21</SUM>
<DE></DE>
<RS>ELibM:1224-2780/4/2/cruz</RS>
<DID>1224-2780/4/2/cruz</DID>
<OI>http://www.emis.de/journals/BJGA/4.2/1.html</OI>
</XREC>

<XREC>
<IDF>Beitr\"age zur Algebra und Geometrie / Contributions to Algebra and Geometry, Vol. 40, No. 1, pp. 203--215</IDF>
<DA>1999</DA>
<IDS>0138-4821</IDS>
<CR>S\"ussmann, Bernd</CR>
<TI>Curve Shortening Flow and the Banchoff-Pohl Inequality on Surfaces of Nonpositive Curvature</TI>
<LA>en</LA>
<SUM>53C65</SUM>
<DE>In this paper the classical Banchoff-Pohl inequality, an isoperimetric inequality for nonsimple closed curves in the Euclidean plane, involving the square of the winding number, is generalized to nonpositively curved surfaces. The proof uses the well-known curve shortening flow.</DE>
<RS>ELibM:0138-4821/40/1/b40h1sue</RS>
<DID>0138-4821/40/1/b40h1sue</DID>
<OI>http://www.emis.de/journals/BAG/vol.40/16.html</OI>
</XREC>

<XREC>
<IDF>Discrete Mathematics & Theoretical Computer Science, Vol. 1, No. 1, pp. 001-016</IDF>
<DA>1997</DA>
<IDS></IDS>
<CR>Schneider, Csaba</CR>
<TI>Computing nilpotent quotients in finitely presented Lie rings</TI>
<LA>en</LA>
<SU>Lie rings nilpotent Lie rings finitely presented Lie rings nilpotent presentation</SU>
<DE>A nilpotent quotient algorithm for finitely presented Lie rings over Z (and Q) is described. The paper studies the graded and non-graded cases separately. The algorithm computes the so-called nilpotent presentation for a finitely presented, nilpotent Lie ring. A nilpotent presentation consists of generators for the abelian group and the products expressed as linear combinations for pairs formed by generators. Using that presentation the word problem is decidable in L. Provided that the Lie ring L is graded, it is possible to determine the canonical presentation for a lower central factor of L. Complexity is studied and it is shown that optimising the presentation is NP-hard. Computational details are provided with examples, timing and some structure theorems obtained from computations. Implementation in C and GAP interface are available.</DE>
<RS>ELibM:/1/1/dm010101</RS>
<DID>/1/1/dm010101</DID>
<OI>http://www.emis.de/journals/DMTCS/0199/1.html</OI>
</XREC>

<XREC>
<IDF>Journal of Convex Analysis, Vol. 6, No. 2, pp. 395--398</IDF>
<DA>1999</DA>
<IDS>0944-6532</IDS>
<CR>Dupin, J. Bair and J. C.</CR>
<TI>The Barrier Cone of a Convex Set and the Closure of the Cover</TI>
<LA>en</LA>
<SUM>52A20</SUM>
<SU>convex set</SU>
<SU>barrier cone</SU>
<SU>recession cone</SU>
<SU>cover</SU>
<SU>polar cone</SU>
<DE>For an arbitrary non-empty closed convex set $A$ in $\mathbb{R}^n$, we prove that the polar of the difference between the barrier cone $\mathbb{B}(A)$ and its interior $\text{int } \mathbb{B} (A)$ coincides with the recession cone $0^+ (\text{cl } \mathbb{G}(A))$ of the closure of the cover $\mathbb{G}(a)$.</DE>
<RS>ELibM:0944-6532/6/2/j170</RS>
<DID>0944-6532/6/2/j170</DID>
<OI>http://www.emis.de/journals/JCA/vol.6_no.2/11.html</OI>
</XREC>
 

<XREC>
<IDF>Journal for Geometry and Graphics\\, Vol. 3, No. 2, pp. 161-176</IDF>
<DA>1999</DA>
<IDS>1433-8157</IDS>
<CR>Pilnikov\'a, Jana</CR>
<CR>Chalmoviansk\'y, Pavel</CR>
<TI>Basis of Quartic Splines over Triangulation</TI>
<LA>en</LA>
<SU>geometric modelling</SU>
<SU>splines</SU>
<SU>triangulation</SU>
<SU>spline basis</SU>
<DE>The modeling of complex shapes usually requires a well-based space of splines. The aim of this work is to give the construction method of such spline space basis over the chosen class of triangulations. This basis has several useful properties - local minimal support, low degree of polynomials. We also present several problems, that arise in lower-degree polynomials.</DE>
<RS>ELibM:1433-8157/3/2/j3h2piln</RS>
<DID>1433-8157/3/2/j3h2piln</DID>
<OI>http://www.emis.de/journals/JGG/3.2/3.html</OI>
</XREC>

<XREC>
<IDF>Journal of Lie Theory, Vol. 9, No. 2, pp. 491--506</IDF>
<DA>1999</DA>
<IDS>0949-5932</IDS>
<CR>Branson, Thomas</CR>
<TI>Spectra of self-gradients on spheres</TI>
<LA>en</LA>
<SUM>58J50</SUM>
<SU>spectral resolution</SU>
<SU>first-order differential operators</SU>
<SU>$n$-sphere</SU>
<SU>Dirac operator</SU>
<SU>Rarita-Schwinger operator</SU>
<DE>We give a general formula for the spectral resolution of a class of first-order differential operators on the sphere $S^n$ which includes, among the most elementary cases, the Dirac and Rarita-Schwinger operators.</DE>
<RS>ELibM:0949-5932/9/2/branslat2e</RS>
<DID>0949-5932/9/2/branslat2e</DID>
<OI>http://www.emis.de/journals/JLT/vol.9_no.2/15.html</OI>
</XREC>
 

<XREC>
<IDF>MATHEMATICA BOHEMICA, Vol. 123, No. 4, pp. 365-369</IDF>
<DA>1998</DA>
<IDS>0862-7959</IDS>
<CR>Janaqi, Stefan</CR>
<CR>F.~Lescure, </CR><CR>M.~Maamoun, </CR>
<CR>H.~Meyniel, </CR>
<TI>Digraphs contractible onto ${}^*\!K_3$.</TI>
<LA>en</LA>
<SUM>05C20</SUM>
<SU>digraph</SU>
<SU>minor</SU>
<SU>contraction</SU>
<DE>We show that any digraph on $n\ge3$ vertices and with not less than $3n-3$ arcs is contractible onto ${}^*\!K_3$</DE>
<RS>ELibM:0862-7959/123/4/jana2796</RS>
<DID>0862-7959/123/4/jana2796</DID>
<OI>http://www.emis.de/journals/MB/123.4/2.html</OI>
</XREC>

<XREC>
<IDF>Publications de l'Institut Math\'ematique (Beograd) (N.S.), Vol. 64 (78), No. 1, pp. 98--106</IDF>
<DA>1998</DA>
<IDS>0350-1302</IDS>
<CR>Stankovi\'c, B.</CR>
<TI>Convergence structures and $S$-asymptotic behaviour of Fourier hyperfunctions</TI>
<LA>en</LA>
<SUM>46F10</SUM>
<DE>Some structural theorems for the convergence in the space of Fourier hyperfunctions are proved and applied to the $S$-asymptotic behaviour of elements in this space.</DE>
<RS>ELibM:0350-1302/64(78)/1/n078p098</RS>
<DID>0350-1302/64(78)/1/n078p098</DID>
<OI>http://www.emis.de/journals/PIMB/078/8.html</OI>
</XREC>

A.3 Guidelines for the Editors contributing to the 
Electronic Library of Mathematics

The following are guidelines for the inclusion of mathematical documents in the Electronic Library of Mathematics (ELibM). This library is a broadly based electronic distribution and archiving network for mathematics. It is provided by the European Mathematical Information Service (EMIS) under the auspices of the European Mathematical Society (EMS). The main URL is http://www.emis.de for EMIS and http://www.emis.de/ELibM.html for ElibM.

The following topics are covered here:

1. ELibM services - what ElibM offers to Journals and where to look for information
2.  Journal structure - how files and directories should be organized
3.  Metadata - how to make sure that users can easily find an article they are looking for
4. Mirroring - how to transmit the journal data to ELibM
5. Journal Production - how ElibM can take care of the whole journal production process for you

A.3.1. ELibM Services

Journals
Proceedings/Collections
Monographs

ELibM offers, in particular, the following services to editors:

- Worldwide distribution and archiving via a system of distributed mirror sites at cooperating institutions (libraries, universities, mathematical societies)
- Production and hosting of electronic journals and collections from electronic files provided by the editors

ElibM services are provided free of charge. User access to ElibM is free as well.

Journal editors will be supported by ELibM shall produce their complete webpages or if they want to produce and host them locally and have ElibM only mirror them. In the first case, editors should read and obey Sections 2-4 of this document, whereas Section 5 describes the process for those journals that want to have their master posting at ELibM.

A.3.2. Journal Structure

ELibM suggests that editors follow these guiding principles:

(1) All files should reside within the directory tree of one directory, the home directory of the journal. The home directory shall contain the home page of the journal, which should be named index.html or index.htm.

Example:  Let us assume here and in the following that /usr/local/httpd/htdocs/EEJ/ is the home directory of EEJ, the "Example ELibM Journal". In this case, the home page is the file /usr/local/httpd/htdocs/EEJ/index.html. The URL of this document would look something like http://your.server.ac.xx/EEJ/index.html. The corresponding EMIS URL will then most probably become http://www.emis.de/journals/EEJ/index.html.

The main directory for HTTP documents assumed above to be named /usr/local/httpd/htdocs/ will be referred to as ~/ in the following, so that ~/EEJ/ becomes the root node of the journal's directory tree.

Caveat: In particular, images and other graphical elements must reside underneath the home directory. This is needed to ensure reliable and easy mirroring of these pages in ELibM.

Example: The journal logo of EEJ may be located in ~/EEJ/images/EEJ.gif. Images must not reside in ~/images/EEJ.gif, because ~/images/ is not a subdirectory of the home directory.

(2) All links to journal pages or graphical elements should be relative links, where relative means that paths should be given relative to the home directory (the root of the journal directory tree).

Example: The home page of EEJ contains a link to the journal logo; this link should look as follows:

<img src=images/EEJ.gif>

because images/ is the path ~/EEJ/images/ relative to the home directory, the root of the journal directory tree.

Note: Absolute linking means that the URL is included in the link, such as in

forbidden: <img src=http://eej.foo.xx/images/EEJ.gif>

Caveat: In particular, all graphical elements have to reside on the same web server. If, for example, the journal's home page has the URL http://eej.foo.xx/, the images must not be under http://www.foo.xx/images/, even if www.foo.xx has the same IP number as eej.foo.xx.

(3) The home page should contain links to tables of contents of all journal volumes resp. issues that are present on the website. The journal volumes (and possibly issues) should be located in subdirectories of the home directory. Preferrably, the volume and issue numbers should also  be the names of the directories. Contents files should be made for either the whole volume or the single issues, not for both.

Example: The contents file for Volume 5, Issue 1 of EEJ is located in ~/EEJ/5/1/index.html, the contents file for Volume 5, Issue 2 is located in 5/2/index.html (relative to the home directory). The journal's home page contains thus the following links:

Volume 5 (1994): <A HREF="5/1/">No. 1</A>, <A HREF="5/2/">No. 2</A>

Alternatively, the whole Volume 5 (all issues) may be contained in one directory, with contents page 5/index.html (relative to the home directory). In this case, the home page would simply contain the link

<A HREF="5/">Volume 5 (1994)</A>

(4) The content page of a volume or issue should contain links to separate abstract pages for each document of the volume (issue). These abstract pages should be in the same directory as the content page and may be named by the document number (e.g., 3.html for the third document in that colume resp. issue), by the first page of the printed article (e.g., 133.html for a document starting with page 133), or in some other way.

Example: The content page for Volume 5, located at 5/index.html, contains the following link:

H. Hermix: < A HREF="5/1.html">On qualitative properties of residual equations on supercomplex subdomains</A>

and this article is the first of this volume.

(5) The abstract page of an article is recommended to contain the following information:
- metadata (see Section 3) (optional)
- authors' names
- title of article (with English translation)
- MSC classifications (optional)
- keywords (optional)
- abstract (optional)
- links to the article document files in DVI, PostScript, or PDF formats

The document files should preferrably reside within the same directory as the abstract page (and thus, in the same directory with all the document files of the given volume resp. issue).

Note: The naming of the document files is not specified. Possible schemes are:
- last names of authors, e.g.: hermix.dvi. hermix.ps, hermix.pdf
- volume, issue, document number information: 5_1_1.dvi, 5_1_1.ps, 5_1_1.pdf

A.3.3. Metadata

The metadata system employed by ELibM is based on the Dublin Core Element Set.

In the case of ELibM-produced journals (see Section 5), metadata are automatically extracted from the content description pages in the journal production process.

If the journal webpages are not themselves produced by ELibM, journal editors should take care to provide the metadata in one of the following ways:
- augmenting the HTML pages with META tags carrying the metadata information
- providing templates for abtsract pages which identify the occurrences of relevant information

A.3.3.1 Augmenting HTML with META tags

Metadata shall be provided in the HTML documenthead of the abstract pages (see Section 2, (5)). They can be conveniently created with the Lund metadata creator.

Example: Here is how the metadata would look for the example of Section 2; the information is embedded in the <HEAD></HEAD> section of the HTML 4.0 code of that page:

<META NAME="DC.Title" CONTENT="On qualitative properties of residual equations on supercomplex subdomains">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#title">

<META NAME="DC.Creator.PersonalName" CONTENT="Hermix, H.">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#creator">

<META NAME="DC.Subject" CONTENT="residual equations">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#subject">

<META NAME="DC.Subject" CONTENT="supercomplex subdomains">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#subject">

<META NAME="DC.Subject" SCHEME="MSC" CONTENT="30G99">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#subject">

<META NAME="DC.Description" CONTENT="We consider some qualitative properties of residual equations on supercomplex subdomains.">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#description">

<META NAME="DC.Publisher" CONTENT="Mathematics Department, Berlin University, Germany">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#publisher">

<META NAME="DC.Date" SCHEME="ISO8601" CONTENT="2000-01-18">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#date">

<META NAME="DC.Type" CONTENT="Text.Article">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#type">

<META NAME="DC.Format" SCHEME="IMT" CONTENT="application/pdf">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#format">
<LINK REL=SCHEMA.imt HREF="http://sunsite.auc.dk/RFC/rfc/rfc2046.html">

<META NAME="DC.Format" SCHEME="IMT" CONTENT="application/postscript">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#format">
<LINK REL=SCHEMA.imt HREF="http://sunsite.auc.dk/RFC/rfc/rfc2046.html">

<META NAME="DC.Format.X-Carrier" CONTENT="file">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#format">

<META NAME="DC.Identifier" SCHEME="ISSN" CONTENT="1234-5678">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#identifier">

<META NAME="DC.Identifier" SCHEME="URL" CONTENT="http://www.berlin.edu/EEJ/5/1/1.html">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#identifier">

<META NAME="DC.Language" SCHEME="ISO639-1" CONTENT="en">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#language">

<META NAME="DC.Rights" CONTENT="Copyright 2000 Berlin University">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#rights">

<META NAME="EULER.Identifier" CONTENT="ELibM Example Journal, Vol. 5, No. 1, pp. 1-23">
<LINK REL=SCHEMA.dc HREF="http://www.lub.lu.se/EULER/partners/metadatacreator.html">

<META NAME="DC.Date.X-MetadataLastModified" SCHEME="ISO8601" CONTENT="2000-01-17">
<LINK REL=SCHEMA.dc HREF="http://purl.org/metadata/dublin_core_elements#date">

A.3.3.2 Providing templates for abstract pages

The idea of templates if that journal editors provide a sample abstract page which identifies the semantic structure of the documents of that journal. The semantic signifiers are called metadata identification tags. Thus, in the place where the author(s) would be listed, the template page would contain instead the metadata identification tag @authors.

The available tags are:

@journaltitle:
@ISSN:
@year:
@volume:
@issue:
@authors:
@affiliation:     (address information for author(s), optional)
@title:
@language:
@pages:
@classifications:     (MSC 2000 classifications)
@keywords:        (optional) English keywords
@abstract:        (optional)
@filename:        (required) without filename extension (dvi,ps,pdf,...)
@other:    (optional) used for other variable information in the page

Example: Assume that the HTML code of our example page is as follows:

<HTML>

<HEAD>
    <TITLE>ELibM Example Journal, Vol. 5, No. 1, pp. 1-23, 2000</TITLE>
</HEAD>

<BODY>
    <B> ELibM Example Journal, Vol. 5, No. 1, pp. 1-23 (2000)</B>

    <HR></HR>

    <CENTER><H1>On qualitative properties of residual equations on supercomplex subdomains</H1>
        <H2>H. Hermix, P. Wooding and R. Slate</H2>
     </CENTER>

    <HR></HR>

    <P><B>Classification (MSC2000):</B> 30G99</P>
    <P><B>Full text of the article:</B></P>
    <P><UL><LI><A HREF="hermix.dvi.gz">Compressed DVI file</A> (26790 Bytes)</UL></P>

    <HR></HR>

    <P><CENTER>
        [<A HREF="2.html">Next Article</A>] [<A HREF="index.html">Contents of this Number</A>]
    </CENTER></P>

    <HR></HR>

    <I><FONT SIZE=-1>Created January 15, 2000
        &copy; 2000 <A href="../../eej.html">ELibM Example Journal</A>
    </FONT></I>

</BODY>
</HTML>

The corresponding template page would look like this:

<HTML>

<HEAD>
    <TITLE>@journaltitle, Vol. @volume, No. @issue, pp. @pages, @year</TITLE>
</HEAD>

<BODY>
    <B> @journaltitle, Vol. @volume, No. @issue, pp. @pages (@year)</B>

    <HR></HR>

    <CENTER><H1>@title</H1>
        <H2>@authors</H2>
     </CENTER>

    <HR></HR>

    <P><B>Classification (MSC2000):</B> @classifications</P>
    <P><B>Full text of the article:</B></P>
    <P><UL><LI><A HREF="@filename.dvi.gz">Compressed DVI file</A> (26790 Bytes)</UL></P>

    <HR></HR>

    <P><CENTER>
        [<A HREF="2.html">Next Article</A>] [<A HREF="index.html">Contents of this Number</A>]
    </CENTER></P>

    <HR></HR>

    <I><FONT SIZE=-1>Created @other
        &copy; @other <A href="../../eej.html">ELibM Example Journal</A>
    </FONT></I>

</BODY>
</HTML>

Caution: The metadata extraction only works if all the variable parts of a page are properly identified by metadata identification tags.

A.3.4. Mirroring

The files on the journal's web server are checked by following recursively all links provided on these pages and checking the file's time stamps. If the time stamp has changed since the last download, then the changed file is retrieved. Otherwise, the local copy already present at ELibM is read and the links on that page are followed. The full texts (all files in formats other than HTML) are never overwritten even if the time stamps have changed. They can only be exchanged by special request. This is done in order to account for the Permanence Principle set forth in the General ELibM Guidelines.

There is also the possibility to upload the files via FTP (File Transfer Protocol) to the EMIS server, or for ELibM to mirror via FTP if the journal's pages are accessible also via this protocol. Details should be negotiated with the technical adminstrators of ELibM.

A.3.5. Journal Production

(a) the single articles
(b) contents and summary information

The single articles should be provided as Postscript, DVI, or TeX files. An additional special contents file should contain table-of-contents information and references to the respective article files (basically the file names). The format of this file is in detail described below. All these files should be put in one empty directory and delivered to EMIS (see below for the details).

The articles

Provision of additional file formats (TeX, DVI, etc.) will be at the discretion of the editors of the corresponding journal. The contents of all versions should essentially agree, although there is no requirement that they all share any particular style.

Filenames should have a common structure. If there are multiple versions (DVI, Postscript, TeX) of the same article, the basename of the file should be the same (e.g. "mumble.dvi", "mumble.ps" etc.).

The contents information

The editor-provided information will also be used for the public display of all articles in the framework of the CAP-EMS Database of current mathematical publications (under construction), and for the faster processing of bibliographic references at Zentralblatt fuer Mathematik.

Furthermore, it will provide the basic means of enabling online delivery of the publications irrespective of protocols and formats. This is an important point when it comes to technology changes.

The contents file is the core for all dissemination activities wrapped around the single articles, and its preparation requires special care.

An example contents file is give in the appendix.
This example was used to create these journal pages automatically.

The file consits of the following parts:

• An identification of the format,
• general information on the issue,
• and entries for the single articles.

@version: EMIS-j-2.0

@journaltitle: (required)
@ISSN:         (optional)
@year:         (required)
@volume:       (required)
@issue:        (required)
@remark:       (optional)

@EOH

@author:          (required)
@affiliation:     (optional)
@title:           (required)
@language:        (required if other than English)
@pages:           (optional)
@classification1: (optional) primary MSC classification
@classification2: (optional) secondary MSC classification(s)
@keywords:        (optional) English keywords
@abstract:        (optional) 
@filename:        (required) without filename extension (dvi,ps,...)

Each article is ended by a line stating

@EOI

Here are some other optional fields that can be used at the editors' discretion:

@contributor:       (e.g.,  the author of an appendix included with the article)
@alternative_title: (subtitle or title translation into English)
@publisher:         (publisher of the journal)
@date:              (date of publication)
@copyright:         (Information about rights held in and over the resource)

Delivery and file transfer mechanism