Metadata

What does "Metadata" mean ?

Metadata simply means "data on data", a concept which provides a better information retrieval mechanism and allows for viewing the 'world of data' from different perspectives. As an example, MetaData of a preprint in mathematics could be:

the name of the author
the title of the preprint
the Mathematical Subject Classification (MSC) which is a very useful Metadatum (65n06 as an example means 'finite difference methods').

How to encode Metadata

Metadata are stored in the head of an HTML document:

<HTML>
<HEAD>
<META NAME="DC.Creator.PersonalName" CONTENT="Grötschel, Martin">
<META NAME="DC.Creator.PersonalName" CONTENT="Dalitz, Wolfgang">
<META NAME="DC.Creator.PersonalName" CONTENT="Winczewski, Vincent">
<META NAME="DC.Date.Created" CONTENT="19981102">
<META NAME="DC.Title" CONTENT="Math-Net: Ein Informationssystem der Mathematik">
<META NAME="DC.Subject.MscPrimary" CONTENT="(SCHEME=MSC91) 94A15">
<META NAME="DC.Type" CONTENT="preprint">
...

In our example we assign keywords to special entries in our classification - which follows the specifications of the Dublin Core classification scheme - such as DC.Creator.Personalname, DC.Title and DC.Date.Created. Then we encode these assignments in a computer-readable form, in this case in HTML 2.0 format and specifically within META tags. All these data are stored in the head of an HTML document. That means they do not appear in our browser. They are only intended for computers and invisible to the human reader.

How to produce Metadata

Though it is very useful to have MSC codes and an abstract for a paper, the only person who is able to produce such metadata for a preprint is the author him-/herself. But since the encoding must be syntactically correct to be readable for computers, it is relatively hard work for someone not being an HTML expert. So what one needs are tools that perform the complicated encoding correctly without bothering the author with HTML details.