Judith Plümer and Roland Schwänzl
Especially in pure mathematics it takes quite some time to get a research article eventually published in a scholarly journal. Two to three years of delay are more an average than an exception. So it became a tradition to send around preprints. In part this custom became organized into preprint series published by mathematical institutes, which also makes for (mild) form of quality control.Following the arrival of the word wide web there was a broad move to electronic dissemination. As practically all of mathematics currently is type set in TeX and converters to PostScript are freely available this was no surprise.
Due to - that's a guess - the developed individualism of mathematicians no such dominating distribution installation like the Los Alamos Server in Physics evolved - at least up till now.
As another ingredient the feeling comes that locally stored electronic versions can be made more easily to follow a potential evolution of a mathematical text.
So as an outgrew of a project on ``Fachinformation'' run by the German Mathematical Society (DMV) the idea of indexing preprints stored on about 50 departmental web servers in Germany was born.
As technical tool the ``Harvest'' Software became proposed to us. Testing started in fall '95 and around the turn of the year 95/96 a first installation was running.
As it turned out completeness is not the issue - that's achieved by setting up a finite almost fixed number of Harvest gatheres - but documentation.
A (pseudo) full text search as is provided by Harvest's evaluation of postscript files - still the most popular format for distribution of mathematics - has it's known limitations.
Obviously it is necessary to add bibliographic information to turn preprint indexing into a quality service.
Another ingredient rather than formal bibliographic information is to include abstracts to present concrete information about the mathematical content of a paper.
A template used by the American Mathematical Society for it's preprint server and Harvest's capability of making use of HTML's Meta tag hinted the direction to go.
The requirements on author provided Meta Information we formulated in early '96. Exactly at the time of the Warwick Meeting in April 96 a technical work group of the DMV was formed. On 15 April 96 the first HTML implementation of the wanted Meta Information was published in the corresponding mailing list. 23 April 96 we first heard of existence of the DublinCore initiative. 30 April 96 we published the first version of a form interface authors can use to produce the Meta information without getting bothered with obscure details of the MetaData set and it's HTML implementation. 12 July we learned about the results of the Warwick meeting and changed the implementation of the MetaData Set to a Warwick container by 1 August 96. As the 15 element DublinCore stabilized in the meantime we moved to a DC implementation by the end of January 97
Since then further development of the preprint index became part of the project ``Informationsdienste für die Mathematik im Internet''.
The following slides show a walk through the installation from production to use of MetaData.
Sample record
Current user interface of the index (Sample query)
Sample result set
Using ``more''
Judith Plümer
Roland Schwänzl
Fb. Mathematik/Informatik
Universität Osnabrück
Albrechtstr. 28
49084 Osnabrück, Germany