Metadata will help to bring structure into the enormous amount of information ,,hidden'' in the Internet. The Dublin Core conventions give a set of well defined semantic elements for metadata. HTML gives the syntactical basis for metadata implementation into documents.

The EuroPhysNet service is a system of Harvest based search engines for information on physics in Europe. As part of the EPRINT project of the Deutsches Forschungsnetz DFN and of a project of the EPS, the European Physical Society, we are improving the PhysDoc broker for information about local documents on servers of the physics departments in Europe. In this connection, the most important problem is the poor semantic and computerreadable structure of most of the distributed documents.

At this point a common set of metadata like DC can help to improve search engines to more intelligent agents. On the other hand the key cornerstone has to be overcome: How to increase the number of documents containing DC metadata?

It is not the idea of DC to have a hand full of experts, who know how to produce metadata, but to let the authors produce meta-information. Nobody else will ever know the document as good as the author himself. Consequently, tools have to be developed, which allow authors to markup the documents with correct and complete DC-meta tags without the author to become an expert of DC before.

Of course the production of metadata is work in itself, but as experienced in physics it can raise the potential of the documents. Metadata become a new kind of value of a document. The tools for metadata production have to be easy to use. This means, that they have to be intuitive and usable without any specific installations on the computer.

Here I want to introduce a first version of such an authoring tool: WUFI - a Webbased-Upload-Form-Interface. WUFI will guide the author to produce metadata and also will upload the document onto the WWW server. Thus, the author doesn't have to know any extra passwords for the WWW server or need to know how to use FTP etc. WUFI is an HTTP form-interface which only requires a Netscape 3.0 browser (or newer). So the author needn't to install any extra software nor have to run any Java applets on his computer. WUFI only uses some JavaScript, but this won't be a problem even on computers with a lack of memory. Various help-texts guide the author through the input forms for metadata and upload information. WUFI will automatically copy the selected document (which of course may contain of several files) from a local disk to a temporary area of the webserver.

To prevent misuse of this upload facility, the author can only upload the document to an internal, temporary file system. In a second step another, privileged user - the so called docmaster - has to finalise the web upload.

The document is automatically listed in the publication list of the work group the author belongs to. It consists of a central HTML-file which contains the metadata, an abstract, and links to the other files. Postscript-files are automatically converted into PDF for easier readability.

During the upload process the document gets an unequivocal identification number. Later the author is able to upgrade the document in a restricted way (as restricted as the institution wants to) or delete it.

We programmed WUFI to increase the number of WWW-publications and use this publication-tool also to guide and force the authors to produce DC-metadata. WUFI was first developed for a prototype being installed at the AWI (Alfred Wegener Institut für Meeres- und Polarforschung).

After a successful betatest, which starts in January 1998 at the Physics Department of the University of Oldenburg, the WUFI software package will be freely available inside the IuK consortium. It will be open for local adaptations, too. Other users ofcourse can use every WUFI installation to generate metadata without document upload.